Test the 503 for a request waiting on a transcode stopped at a deadline

A request that waited for another request's transcode, stopped at that
request's deadline, must get ErrProcessingTimedOut, which both image
routes answer 503. Declares that error so the tests build. Also tests
that Process, past its deadline before it starts, never reads its input.

Model: opus-5-5

Dockerfile

@@ -32,10 +32,13 @@ RUN script/bootstrap --cgo
 COPY . .
 
 # Without -v first; on a failure, again with -v for the details, and
-# the step fails even if the second run passes.
-RUN go test -count=1 -timeout 90s -race -cover ./... || \
+# the step fails even if the second run passes. -parallel 4: by default
+# a package runs as many of its tests at once as the host has CPUs, and
+# on a busy host they then wait so long to be scheduled that a test
+# that times a request can see it return a second late.
+RUN go test -count=1 -timeout 90s -race -parallel 4 -cover ./... || \
     { echo "--- Rerunning with -v for details ---"; \
-      go test -count=1 -timeout 90s -race -v ./...; exit 1; }
+      go test -count=1 -timeout 90s -race -parallel 4 -v ./...; exit 1; }
 
 # Build stage. Nothing is wanted from the two phases above: these copies
 # make BuildKit build them first, so this stage runs only when lint and

README.md

@@ -26,6 +26,11 @@ make docker
 docker run -p 8080:8080 -e PIXA_SIGNING_KEY="$(openssl rand -base64 32)" pixa:latest
 ```
 
+The example config allowlists no host (`allowlist_hosts`), so pixa serves an
+image only for a signed or encrypted URL. Every image on an allowlisted host is
+served to anyone without a signature, at any size and in any format, so
+allowlist only hosts you control, where nobody else can put images.
+
 A container takes its settings from environment variables (see Configuration
 below for the list). Only `PIXA_SIGNING_KEY` is required; if it is unset the
 container exits at startup naming the variable. Everything else has a built-in
@@ -46,9 +51,9 @@ everything in this list without further settings. The reverse proxy must:
   builds encrypted URLs from `Host`, and checks `Referer` against
   `referer_blocklist`;
 - set `X-Forwarded-For` to the client's address, with `trusted_proxies` set to
-  the address pixa sees the proxy's requests come from, so the login limit
-  counts each client by its own address (see `trusted_proxies` under
-  Configuration);
+  the address pixa sees the proxy's requests come from, so the login limit and
+  `image_requests_per_minute` count each client by its own address (see
+  `trusted_proxies` under Configuration);
 - wait for pixa's answer for at least `downstream_timeout` (default `60s`), the
   longest pixa takes to fetch, convert and send an image.
 
@@ -58,8 +63,9 @@ password there.
 
 pixa does the rest itself: it checks signatures and encrypted URLs, applies the
 allowlist, refuses upstream hosts with private or local addresses, limits login
-attempts, upstream response size and image dimensions, and sends the security
-headers, `Strict-Transport-Security` included, with every response.
+attempts, each client's image requests, upstream response size and image
+dimensions, and sends the security headers, `Strict-Transport-Security`
+included, with every response.
 
 The state directory (`state_dir`, `/var/lib/pixa` in the container) holds the
 database and the disk cache:
@@ -83,10 +89,12 @@ which answers 200 whenever pixa is running, in maintenance mode too (see
 `maintenance_mode`).
 
 On SIGTERM or SIGINT pixa stops accepting connections, gives the requests in
-progress and the images being processed 5 seconds to finish, and exits: with 0,
-or with 1 when images were still being processed after those 5 seconds or
-another part of pixa failed to stop. A request not finished by then is cut off.
-`docker stop` waits 10 seconds before it kills the container.
+progress and the images being processed 5 seconds to finish, then writes to the
+database the counts of cache hits, misses, fetches and conversions that requests
+could not write by their deadline, and exits: with 0, or with 1 when images were
+still being processed after those 5 seconds, some of those counts could not be
+written, or another part of pixa failed to stop. A request not finished by then
+is cut off. `docker stop` waits 10 seconds before it kills the container.
 
 Outside Docker, pixa needs libvips (the image has 8.16) and libheif to run, as
 it uses libvips through CGO. pixad does not start unless libvips has its JPEG XL
@@ -112,7 +120,9 @@ What the [upaas](https://git.eeqj.de/sneak/upaas) app for pixa needs:
     - `PIXA_SIGNING_KEY` (required): secret for signed and encrypted URLs and
       login, 32+ characters, for example from `openssl rand -base64 32`
     - `PIXA_ALLOWLIST_HOSTS`: upstream hosts served without a signature,
-      comma-separated
+      comma-separated; default none. Every image on a listed host is served to
+      anyone without a signature, so list only hosts you control, where nobody
+      else can put images.
     - `PIXA_CACHE_MAX_BYTES`: disk cache limit in bytes; `0` disables it;
       default 75% of (free space + what the cache holds)
     - the rest are in the table under Configuration below
@@ -186,7 +196,10 @@ path under `/v1/` answers 200, in maintenance mode too.
   `Referer` names a host in `referer_blocklist`, checked before the signature,
   the cache and the upstream fetch; 403 when the upstream host, or a host it
   redirects to, is `localhost`, ends in `.localhost` or `.local`, or has an
-  address in a blocked network (see `blocked_networks`); 502 when the upstream
+  address in a blocked network (see `blocked_networks`); 429 over
+  `image_requests_per_minute` (below); 415 when the upstream answers with a
+  `Content-Type` that is not one of the source formats (below), or with an image
+  whose first bytes libvips reads as another format; 502 when the upstream
   answered with an error status, and for 5 minutes after that for the same
   source URL; 503 when pixa is busy or in maintenance mode; 500 for any other
   failure.
@@ -197,7 +210,7 @@ path under `/v1/` answers 200, in maintenance mode too.
   expired; 504 when the upstream has not sent its response headers within
   `upstream_fetch_timeout`, but 500 when that time runs out while the image
   itself is still arriving; 400 for an `Accept` header that is not valid, and
-  406, 403, 502, 503 and 500, as for `/v1/image/`.
+  406, 403, 415, 429, 502, 503 and 500, as for `/v1/image/`.
 - `GET /robots.txt` — asks every crawler to stay away (`Disallow: /`). Needs:
   nothing. Answers: 200.
 - `GET /.well-known/healthcheck.json` — JSON with `status` (`ok`), `now`,
@@ -257,9 +270,17 @@ once, is refused with 400.
   both optional (values under Signature Specification). Both are part of what is
   cached, so each value of either is a separate cached image.
 
-The source image may be JPEG, PNG, GIF, WebP, AVIF, JPEG XL or SVG. The
-upstream's `Content-Type` must name its format, and the image's first bytes must
-match it.
+A size with one side `0` has that side scaled to keep the source's proportions,
+and is refused with 400 when that side would come out above 8192, the most
+either side may be; so is a size with `fit=outside`, which scales the image to
+cover the size and so past it on one side, when that side would come out
+above 8192.
+
+The source image may be JPEG, PNG, GIF, WebP, AVIF or JPEG XL. The upstream's
+`Content-Type` must name its format, and the image's first bytes must match it.
+An upstream answering with any other `Content-Type`, SVG included, is refused
+with 415, and so is an image whose first bytes libvips reads as another format,
+such as an SVG whose first bytes pass the check for AVIF.
 
 With the format `auto`, pixa chooses the format for each request from its
 `Accept` header, in this order:
@@ -292,9 +313,23 @@ cache at once, they share one upstream fetch (or one read of the cached source)
 and one transcode: the first request does the work, and the others wait for its
 image or its error, holding no upstream connection or processing slot of their
 own. A waiting request stops waiting when its own client goes away. The work
-goes on for the others even if the first request's client goes away, until that
-request's `downstream_timeout` ends. The shared fetch sends the first request's
-ID upstream, and the lines logged for the fetch and the transcode carry that ID.
+goes on for the others even if the first request's client goes away. The shared
+fetch sends the first request's ID upstream, and the lines logged for the fetch
+and the transcode carry that ID.
+
+A transcode whose request's `downstream_timeout` has ended (for a shared one,
+the first request's) stops when it gets its processing slot or before it begins
+encoding, and gives the slot back. libvips cannot stop an encode once it has
+begun, so an encode keeps its slot until it finishes, which can be after
+`downstream_timeout`. The limit of 8192 on each side of the source and of the
+output bounds how long that takes. With one libvips thread, as pixad runs it,
+encoding 8192x8192 random pixels takes up to about 1 second as JPEG, 15 as PNG,
+16 as GIF, 63 as WebP, 110 as AVIF and 96 as JPEG XL, the slowest of 8 or 16
+bits per sample, with or without an alpha channel. WebP, AVIF and JPEG XL are
+that slow only with an alpha channel: 8-bit RGB takes about 44, 53 and 28
+seconds. Decoding such a source adds up to about 11 seconds. A WebP, AVIF or
+JPEG XL encode can therefore hold its slot past the default `downstream_timeout`
+of 60 seconds.
 
 The login form (`POST /`) is limited to 5 attempts per minute per client
 address, counting an IPv6 client by its /64; an attempt over the limit is
@@ -313,6 +348,15 @@ its own `X-Forwarded-For`, whether it connects directly or through the proxy,
 because its own address is trusted too. Setting `trusted_proxies` to only the
 address pixa sees for requests that come through the proxy closes this.
 
+One client may send `image_requests_per_minute` requests per minute
+(default 600) to the image routes, `/v1/image/` and `/v1/e/` together, `GET` and
+`HEAD`, cache hits included; a request over the limit is refused with 429 and a
+`Retry-After` header. Clients are counted as for the login limit. A browser's
+CORS preflight is not counted. Behind a CDN or proxy whose addresses are not in
+`trusted_proxies`, all requests arriving from one proxy or CDN address count as
+one client, so such a deployment needs a higher limit, or `0`, which turns the
+limit off.
+
 ### Encrypted URLs
 
 An encrypted URL is an image URL made on pixa's own web page by someone who
@@ -328,7 +372,8 @@ nor change what it asks for.
    (`POST /generate`). The format is JPEG XL unless another is chosen; a form
    sent with an empty format, or none, also makes a JPEG XL URL. Width and
    height both empty or `0` keep the original size; if only one of them is empty
-   or `0`, that side is scaled to keep the image's proportions.
+   or `0`, that side is scaled to keep the image's proportions. A URL whose
+   scaled side would come out above 8192 answers 400.
 3. The page shows the URL, `https://<host>/v1/e/<token>/img.<format>`, and when
    it expires. `<host>` is the host the page was opened on, and the URL starts
    with `http` instead while `debug` is on. The name after the token is ignored
@@ -388,8 +433,11 @@ HMAC-SHA256(secret, "host:path:query:width:height:format:expiration:quality:fit"
 Where:
 
 - `host` — source origin hostname (e.g. `cdn.example.com`)
-- `path` — source path (e.g. `/photos/cat.jpg`)
-- `query` — source query string, empty string if none
+- `path` — source path (e.g. `/photos/cat.jpg`), with each `%` written as `%25`
+  and each `:` as `%3A`, so `/photos/a:b.jpg` is signed as `/photos/a%3Ab.jpg`
+- `query` — source query string, empty string if none, with `%` and `:` written
+  as in `path`; neither then holds a `:`, so path `/a:b` with no query and path
+  `/a` with query `b:` are signed differently
 - `width` — requested width in pixels, `0` for original
 - `height` — requested height in pixels, `0` for original
 - `format` — output format, one of those listed under Routes, with `original`
@@ -482,6 +530,7 @@ tell. With no file, pixa uses the environment and the defaults.
 | `PIXA_UPSTREAM_CONNECTIONS_PER_HOST` | `upstream_connections_per_host` | Concurrent connections per upstream host; default `20`                       |
 | `PIXA_UPSTREAM_CONNECTIONS`          | `upstream_connections`          | Concurrent connections to all upstream hosts together; default `64`          |
 | `PIXA_MAX_CONCURRENT_PROCESSING`     | `max_concurrent_processing`     | Images processed at once; default the number of CPUs                         |
+| `PIXA_IMAGE_REQUESTS_PER_MINUTE`     | `image_requests_per_minute`     | Image requests per client per minute; `0` turns it off; default `600`        |
 | `PIXA_UPSTREAM_FETCH_TIMEOUT`        | `upstream_fetch_timeout`        | Time allowed for one fetch from an upstream host; default `30s`              |
 | `PIXA_UPSTREAM_MAX_RESPONSE_SIZE`    | `upstream_max_response_size`    | Largest upstream response accepted, in bytes; default 50 MiB                 |
 | `PIXA_DOWNSTREAM_TIMEOUT`            | `downstream_timeout`            | Time allowed for answering one client request; default `60s`                 |
@@ -531,13 +580,13 @@ Key settings in more detail:
   explicit list replaces the default. An invalid CIDR aborts startup. Set this
   to the address pixa sees for requests that come through your proxy, such as
   `172.17.0.1/32`, when the defaults do not cover it, or to trust nothing else
-  (see the login limit under Routes). For a proxy on the Docker host that
-  connects to pixa over `127.0.0.1`, that address is the gateway of the
-  container's Docker network (`172.17.0.1` on the default bridge), not the
-  proxy's own address; a proxy that connects through another of the host's
-  addresses is seen with that address. To be sure which address it is, set this
-  to `[]` (or `PIXA_TRUSTED_PROXIES` to empty), send a request through the
-  proxy, and read `remoteIP` in pixa's request log line for it
+  (see the login limit and `image_requests_per_minute` under Routes). For a
+  proxy on the Docker host that connects to pixa over `127.0.0.1`, that address
+  is the gateway of the container's Docker network (`172.17.0.1` on the default
+  bridge), not the proxy's own address; a proxy that connects through another of
+  the host's addresses is seen with that address. To be sure which address it
+  is, set this to `[]` (or `PIXA_TRUSTED_PROXIES` to empty), send a request
+  through the proxy, and read `remoteIP` in pixa's request log line for it
 - `upstream_fetch_timeout` — time allowed for one fetch from an upstream host,
   as a duration such as `30s` (the default) or `2m`
 - `upstream_max_response_size` — largest upstream response accepted, in bytes;
@@ -570,6 +619,12 @@ Key settings in more detail:
   container's CPU limit. A request that finds all of them in use waits up to 10
   seconds for one to free up; if none does, and `downstream_timeout` has not
   ended first, it is answered 503 the same way
+- `image_requests_per_minute` — how many requests one client may send to the
+  image routes per minute (see the limit on them under Routes); default `600`.
+  `0` turns the limit off, and a negative or non-integer value aborts startup.
+  Behind a CDN or proxy whose addresses are not in `trusted_proxies`, all
+  requests arriving from one proxy or CDN address count as one client, so raise
+  the limit there, or set `0`
 - `maintenance_mode` — while `true`, the image routes (`/v1/image/` and
   `/v1/e/`) answer every request for an image with 503, a `Retry-After` header
   and a JSON error body. The health check (`/.well-known/healthcheck.json`)

TODO.md

@@ -26,10 +26,108 @@ resolved. The disk cache is now size-bounded with LRU eviction
 
 # Next Step
 
-P2: security: per-IP rate limiting on the image routes
+P2: security: per-origin rate limiting
 
 # Completed Steps
 
+- 2026-10-15 a `:` in the source path or query no longer moves where the path
+  ends and the query begins (closes #247): the signed data and the cache key
+  joined host, path and query with `:` and no escaping, so a URL signed for
+  `https://h/a:b` verified for `https://h/a?b:`; the two also shared a cache
+  key, and whichever was fetched first was served for both. In both, the path
+  and the query now have each `%` written as `%25` and each `:` as `%3A`; a path
+  and query holding neither are signed and keyed as before. Signature
+  Specification in `README.md` says so.
+- 2026-10-15 a transcode stops once its request's `downstream_timeout` has
+  ended, unless it has begun encoding (closes #243): the image processor looked
+  at its context only while it waited for a processing slot, so an image was
+  decoded and encoded, holding the slot, however long after its request's
+  deadline that took. `Process` now returns the context's error, and gives the
+  slot back, when the context has ended by the time it has the slot or before it
+  encodes. libvips cannot stop an encode once it has begun, so that encode keeps
+  its slot until it finishes; the limit of 8192 on each side of the source and
+  of the output bounds how long that takes. With one libvips thread, 8192x8192
+  random pixels take up to about 1 second to encode as JPEG, 15 as PNG, 16 as
+  GIF, 63 as WebP, 110 as AVIF and 96 as JPEG XL, the last three with an alpha
+  channel, plus up to about 11 seconds to decode the source; `README.md` gives
+  these times and no longer says the work stops when `downstream_timeout` ends.
+  The settings of the WebP, AVIF and JPEG XL encoders, which can hold a slot
+  past the default `downstream_timeout` of 60 seconds, are unchanged.
+- 2026-10-15 the example config allowlists no host (closes #249): it allowlisted
+  `github.com` and `user-images.githubusercontent.com`, where anyone can put an
+  image, so a deployment set up as Getting Started says served those images to
+  anyone without a signature, at any size and in any format.
+  `configs/config.example.yml` now leaves `allowlist_hosts` out, with one
+  commented entry under `example.com`. Getting Started and "Running under upaas"
+  in `README.md` say that every image on an allowlisted host is served without a
+  signature, so an allowlisted host must be one the operator controls.
+- 2026-10-15 a chunked upstream body of exactly `upstream_max_response_size`
+  bytes is accepted (closes #248): once the limit was used up, the fetcher's
+  next read refused the body without looking for its end. Go's HTTP client
+  returns the end of a body sent with a `Content-Length` together with its last
+  bytes, but the end of a chunked body can come in a later read, so such a body
+  was refused after it had been downloaded whole, and the image routes
+  answered 500. Once the limit is used up, the fetcher now reads one more byte:
+  if the body has ended it is accepted, and if there is more it is refused with
+  `ErrResponseTooLarge`, as before.
+- 2026-10-13 `fit=outside` no longer makes an output above 8192 on a side
+  (closes #257): it scales the image to cover the requested size, and the side
+  scaled past that size was held to no maximum, so `1x8192?fit=outside` of a
+  3000x600 source was resized to 40960x8192 and encoded. The image processor now
+  refuses such a size with `ErrOutputTooLarge` before resizing, and both image
+  routes answer it 400, as for a size with one side `0`. The other fit modes
+  make the requested size or one within it, so none can go above 8192.
+- 2026-10-13 a size with one side `0` no longer makes an output above 8192 on a
+  side (closes #241): that side is computed from the source's proportions, and
+  nothing held it to the 8192 a requested side is held to, so `0x8192` of a
+  3000x600 source was resized to 40960x8192 and encoded. The image processor now
+  refuses such a size with `ErrOutputTooLarge` once it knows the source's
+  dimensions, before resizing, and both image routes answer it 400 with the JSON
+  error body. The refusal is not cached as a failed fetch, so the same source is
+  still served at other sizes. `imgcache.MaxDimension` is now the image
+  processor's `MaxOutputDimension`, so the two cannot differ.
+- 2026-10-13 the unused cache lifetime is removed (closes #69):
+  `CacheConfig.CacheTTL`, which pixad set to `DefaultCacheTTL` (24 hours), was
+  never read, so cached images never expired. The field and the constant are
+  removed. Cached source and transformed images stay until eviction removes them
+  to keep the cache within `cache_max_bytes`.
+- 2026-10-13 SVG sources are refused, and `orig` is never saved as JPEG in place
+  of the source's own format (closes #68): the upstream fetch no longer accepts
+  `image/svg+xml`, which libvips would render. Both image routes answer 415 for
+  an upstream `Content-Type` that is not one of the source formats, SVG
+  included, where they answered 500. Before decoding, the image processor
+  refuses with `ErrUnsupportedInputFormat`, also answered 415, any input that
+  govips' `DetermineImageType`, which picks the libvips loader from the first
+  bytes, does not name as a source format. That catches an SVG whose first bytes
+  pass the first-bytes check for another type, as an XML comment can make them
+  do for AVIF; an SVG whose first bytes do not pass that check answers 500, as
+  before, and so does a damaged image in a source format, whose first bytes
+  `DetermineImageType` names. `orig` is always the source's own format, where a
+  source the processor could not name was saved as JPEG. `README.md` no longer
+  lists SVG. The SVG detection in `internal/magic` stays: the request path does
+  not call it, and it goes with the unused code of
+  https://git.eeqj.de/sneak/pixa/issues/168.
+- 2026-10-12 per-client limit on the image routes (closes #236): the new setting
+  `image_requests_per_minute` (`PIXA_IMAGE_REQUESTS_PER_MINUTE`), default 600,
+  is how many `GET` and `HEAD` requests one client may send to `/v1/image/` and
+  `/v1/e/` together per minute, cache hits included; the next is answered 429
+  with `Retry-After`. It uses the login limit's `RateLimit` middleware, so
+  clients are told apart the same way, and runs after the CORS middleware, so a
+  preflight is not counted. `0` turns the limit off; a negative or non-integer
+  value aborts startup.
+- 2026-10-08 a request past its deadline no longer waits for the database to
+  count it (closes #224). On a busy host, `TestService_Get_ReturnsByItsDeadline`
+  failed because its request, past its deadline, still waited for its miss count
+  to be written, a write with no deadline, which in pixad waits for the one
+  database connection every request shares. Each count write now keeps the
+  request's deadline but not its cancellation. A count not written by then is
+  kept in memory, where `Stats` sees it, and one goroutine of the cache writes
+  those counts, one write at a time, and once more at shutdown before the
+  database closes. The test phase of the `Dockerfile` also runs at most 4 tests
+  of a package at once (`go test -parallel 4`), where it ran as many as the host
+  has CPUs: on a busy host they then wait so long to be scheduled that a test
+  that times a request can fail. `-p`, how many packages are tested at once, is
+  unchanged, as capping it also slows compiling.
 - 2026-10-08 every output format is saved with settings pixa sets on purpose
   (closes #232): each format has its own govips export, as JPEG XL does, in
   place of govips' generic `Export`, which sent libvips a zero for some settings
@@ -418,11 +516,11 @@ P2: security: per-IP rate limiting on the image routes
   cancellation but keeps its deadline, and the others wait for its image or
   error holding no upstream connection or processing slot, and stop waiting when
   their own context ends; the request doing the processing waits for it even
-  then, up to its deadline; a request whose context has already ended starts
-  nothing; each request counts one miss, and the processing counts its fetch and
-  transcode once; a panic while processing is reported to Sentry when
-  `sentry_dsn` is set and becomes an error for every waiting request instead of
-  stopping pixad; documented in `README.md`.
+  then, until it ends, which can be after its deadline; a request whose context
+  has already ended starts nothing; each request counts one miss, and the
+  processing counts its fetch and transcode once; a panic while processing is
+  reported to Sentry when `sentry_dsn` is set and becomes an error for every
+  waiting request instead of stopping pixad; documented in `README.md`.
 - 2026-09-29 only the image routes send CORS headers (closes #98): the CORS
   middleware, with the `access_control_allow_origin` origin, moved from the
   router root onto a `/v1` subrouter holding `/v1/image/` and `/v1/e/`, where it
@@ -736,8 +834,6 @@ P2: security: per-IP rate limiting on the image routes
 
 # Future Steps
 
-- P2: security
-    - per-origin rate limiting
 - P2: HTTP response handling
     - Last-Modified headers
 - P2: configuration

configs/config.example.yml

@@ -45,15 +45,14 @@ state_dir: ./data
 signing_key: "CHANGE_ME_generate_with_openssl_rand_base64_32"
 
 # Hosts that don't require signatures (default: none)
+# Every image on an allowlisted host is served to anyone without a
+# signature, at any size and in any format, so list only hosts you
+# control, where nobody else can put images.
 # Use "." prefix for wildcard subdomain matching (e.g., ".example.com" matches "cdn.example.com")
 # An entry that is neither a host name nor an IP address (IPv6 without
 # brackets), such as one with a port or a "*." wildcard, aborts startup.
-allowlist_hosts:
-  - s3.sneak.cloud
-  - static.sneak.cloud
-  - sneak.berlin
-  - github.com
-  - user-images.githubusercontent.com
+# allowlist_hosts:
+#   - images.example.com
 
 # Hosts whose pages may not show pixa's images, written as for
 # allowlist_hosts. A request to /v1/image/ or /v1/e/ whose Referer header
@@ -115,6 +114,14 @@ upstream_connections: 64
 # frees up it is answered 503, unless downstream_timeout has ended first.
 # max_concurrent_processing: 4
 
+# How many requests to the image routes (/v1/image/ and /v1/e/) one client
+# may send per minute; a request over the limit is answered 429 with a
+# Retry-After header. 0 turns the limit off. Clients are told apart as for
+# the login limit, so behind a CDN or proxy whose addresses are not in
+# trusted_proxies all requests arriving from one proxy or CDN address count
+# as one client: raise the limit there, or set 0. (default: 600)
+image_requests_per_minute: 600
+
 # Time allowed for one fetch from an upstream host (default: 30s)
 upstream_fetch_timeout: 30s
 

internal/config/config.go

@@ -34,6 +34,7 @@ const (
 	DefaultUpstreamFetchTimeout       = 30 * time.Second
 	DefaultUpstreamMaxResponseSize    = 50 << 20 // 50 MiB
 	DefaultDownstreamTimeout          = 60 * time.Second
+	DefaultImageRequestsPerMinute     = 600
 )
 
 // Configuration key names.
@@ -61,6 +62,7 @@ const (
 	keyUpstreamFetchTimeout       = "upstream_fetch_timeout"
 	keyUpstreamMaxResponseSize    = "upstream_max_response_size"
 	keyDownstreamTimeout          = "downstream_timeout"
+	keyImageRequestsPerMinute     = "image_requests_per_minute"
 )
 
 // placeholderSigningKey is the dummy signing_key shipped in
@@ -158,6 +160,11 @@ type Config struct {
 	// of the per-request timeout middleware.
 	DownstreamTimeout time.Duration
 
+	// ImageRequestsPerMinute is how many requests one client may make to
+	// the image routes per minute; the next is refused with 429 Too Many
+	// Requests. 0 turns the limit off.
+	ImageRequestsPerMinute int
+
 	// BlockedNetworks are operator-supplied CIDR ranges to refuse in
 	// addition to the built-in SSRF blocklist. Enforced by the upstream
 	// fetcher's dialer; the built-in ranges always apply.
@@ -302,6 +309,8 @@ func newFromSmartConfig(sc *smartconfig.Config) (*Config, error) {
 			keyAccessControlAllowOrigin, DefaultAccessControlAllowOrigin),
 		DownstreamTimeout: loader.durationVal(
 			keyDownstreamTimeout, DefaultDownstreamTimeout),
+		ImageRequestsPerMinute: loader.intVal(
+			keyImageRequestsPerMinute, DefaultImageRequestsPerMinute),
 		CacheMaxBytes:    loader.int64Val(keyCacheMaxBytes, 0),
 		BlockedNetworks:  blockedNetworks,
 		TrustedProxies:   trustedProxies,
@@ -311,9 +320,7 @@ func newFromSmartConfig(sc *smartconfig.Config) (*Config, error) {
 	// The default for an omitted cache_max_bytes is worked out when
 	// the cache opens; here we only record whether the operator set
 	// the key explicitly.
-	if _, present := lookupValue(sc, keyCacheMaxBytes); present {
-		c.CacheMaxBytesExplicit = true
-	}
+	_, c.CacheMaxBytesExplicit = lookupValue(sc, keyCacheMaxBytes)
 
 	// Build DBURL from StateDir if not explicitly set. The derived URL
 	// is a default: it applies only when db_url is omitted, never to an
@@ -426,7 +433,7 @@ func isKnownConfigKey(key string) bool {
 		keyMaxConcurrentProcessing, keyCacheMaxBytes, keyBlockedNetworks,
 		keyTrustedProxies, keyAccessControlAllowOrigin, keyUpstreamFetchTimeout,
 		keyUpstreamMaxResponseSize, keyDownstreamTimeout, keyRefererBlocklist,
-		"env":
+		keyImageRequestsPerMinute, "env":
 		return true
 	}
 
@@ -461,6 +468,7 @@ func envVarNames() map[string]string {
 		keyUpstreamFetchTimeout:       "PIXA_UPSTREAM_FETCH_TIMEOUT",
 		keyUpstreamMaxResponseSize:    "PIXA_UPSTREAM_MAX_RESPONSE_SIZE",
 		keyDownstreamTimeout:          "PIXA_DOWNSTREAM_TIMEOUT",
+		keyImageRequestsPerMinute:     "PIXA_IMAGE_REQUESTS_PER_MINUTE",
 	}
 }
 
@@ -613,6 +621,14 @@ func (c *Config) validate() error {
 			settingName(keyCacheMaxBytes), c.CacheMaxBytes, errMustNotBeNegative)
 	}
 
+	// Zero is valid (it turns the limit off); only negative values are
+	// rejected.
+	if c.ImageRequestsPerMinute < 0 {
+		return fmt.Errorf("%s: value %d %w",
+			settingName(keyImageRequestsPerMinute), c.ImageRequestsPerMinute,
+			errMustNotBeNegative)
+	}
+
 	err = c.validateUpstreamMaxResponseSize()
 	if err != nil {
 		return err
@@ -625,12 +641,9 @@ func (c *Config) validate() error {
 		}
 	}
 
-	if c.SentryDSN != "" {
-		parsed, err := url.Parse(c.SentryDSN)
-		if err != nil || parsed.Scheme == "" || parsed.Host == "" {
-			return fmt.Errorf("%s: value %q is %w",
-				settingName(keySentryDSN), c.SentryDSN, errNotAValidURL)
-		}
+	err = c.validateSentryDSN()
+	if err != nil {
+		return err
 	}
 
 	if (c.MetricsUsername == "") != (c.MetricsPassword == "") {
@@ -642,6 +655,22 @@ func (c *Config) validate() error {
 	return c.validateAccessControlAllowOrigin()
 }
 
+// validateSentryDSN checks that sentry_dsn, when set, is a URL with a scheme
+// and a host.
+func (c *Config) validateSentryDSN() error {
+	if c.SentryDSN == "" {
+		return nil
+	}
+
+	parsed, err := url.Parse(c.SentryDSN)
+	if err != nil || parsed.Scheme == "" || parsed.Host == "" {
+		return fmt.Errorf("%s: value %q is %w",
+			settingName(keySentryDSN), c.SentryDSN, errNotAValidURL)
+	}
+
+	return nil
+}
+
 // validateUpstreamMaxResponseSize checks that upstream_max_response_size
 // is from 1 byte to 1 GiB. An upstream response is read whole into
 // memory, and the image processor reads one byte past this limit, which

internal/config/config_internal_test.go

@@ -92,6 +92,32 @@ func TestGetStringSlice_Empty(t *testing.T) {
 	}
 }
 
+// TestExampleConfigAllowlistsNoHost loads configs/config.example.yml
+// with a real signing key, as Getting Started in README.md has it copied
+// and edited, and checks that it allowlists no host: every image on an
+// allowlisted host is served to anyone without a signature.
+func TestExampleConfigAllowlistsNoHost(t *testing.T) {
+	t.Setenv("PIXA_SIGNING_KEY", validTestSigningKey)
+
+	examplePath := filepath.Join("..", "..", "configs", "config.example.yml")
+
+	sc, err := loadTestConfig(examplePath)
+	if err != nil {
+		t.Fatalf("failed to load %s: %v", examplePath, err)
+	}
+
+	c, err := newFromSmartConfig(sc)
+	if err != nil {
+		t.Fatalf("%s is not a valid config: %v", examplePath, err)
+	}
+
+	t.Logf("allowlist_hosts in %s: %v", examplePath, c.AllowlistHosts)
+
+	if len(c.AllowlistHosts) != 0 {
+		t.Errorf("%s allowlists %v, want no host", examplePath, c.AllowlistHosts)
+	}
+}
+
 // loadTestConfig is a helper to load a config file for testing.
 func loadTestConfig(path string) (*smartconfig.Config, error) {
 	return smartconfig.NewFromConfigPath(path)

internal/config/env_internal_test.go

@@ -70,6 +70,7 @@ func TestEnvironmentSetsEveryKey(t *testing.T) {
 	t.Setenv("PIXA_UPSTREAM_CONNECTIONS_PER_HOST", "5")
 	t.Setenv("PIXA_UPSTREAM_CONNECTIONS", "10")
 	t.Setenv("PIXA_MAX_CONCURRENT_PROCESSING", "3")
+	t.Setenv("PIXA_IMAGE_REQUESTS_PER_MINUTE", "30")
 	t.Setenv("PIXA_CACHE_MAX_BYTES", "1024")
 	t.Setenv("PIXA_BLOCKED_NETWORKS", "203.0.113.0/24")
 	t.Setenv("PIXA_TRUSTED_PROXIES", "192.0.2.0/24")
@@ -99,6 +100,7 @@ func TestEnvironmentSetsEveryKey(t *testing.T) {
 		UpstreamConnectionsPerHost: 5,
 		UpstreamConnections:        10,
 		MaxConcurrentProcessing:    3,
+		ImageRequestsPerMinute:     30,
 		CacheMaxBytes:              1024,
 		CacheMaxBytesExplicit:      true,
 		BlockedNetworks:            []netip.Prefix{netip.MustParsePrefix("203.0.113.0/24")},

internal/config/image_requests_per_minute_internal_test.go

@@ -0,0 +1,107 @@
+package config
+
+import "testing"
+
+// testImageRequestsPerMinuteVar is the variable that sets
+// image_requests_per_minute.
+const testImageRequestsPerMinuteVar = "PIXA_IMAGE_REQUESTS_PER_MINUTE"
+
+// TestImageRequestsPerMinuteValues checks that an omitted
+// image_requests_per_minute is 600, and that 0, which turns the limit off,
+// and any larger whole number are used as given.
+func TestImageRequestsPerMinuteValues(t *testing.T) {
+	t.Parallel()
+
+	c, err := configFromYAML(t, signingKeyLine)
+	if err != nil {
+		t.Fatalf("minimal config should be valid, got error: %v", err)
+	}
+
+	if c.ImageRequestsPerMinute != 600 {
+		t.Errorf("ImageRequestsPerMinute = %d, want 600", c.ImageRequestsPerMinute)
+	}
+
+	explicit := []struct {
+		value string
+		want  int
+	}{
+		{"0", 0},
+		{"30", 30},
+	}
+
+	for _, tc := range explicit {
+		c, err := configFromYAML(t,
+			signingKeyLine+"image_requests_per_minute: "+tc.value+"\n")
+		if err != nil {
+			t.Fatalf("image_requests_per_minute: %s should load, got error: %v",
+				tc.value, err)
+		}
+
+		if c.ImageRequestsPerMinute != tc.want {
+			t.Errorf("ImageRequestsPerMinute = %d, want %d",
+				c.ImageRequestsPerMinute, tc.want)
+		}
+	}
+}
+
+// TestInvalidImageRequestsPerMinuteAbortsStartup checks that a value that is
+// negative, not a whole number, or null aborts startup naming the key and
+// the value, and the variable too where the value could have come from it.
+func TestInvalidImageRequestsPerMinuteAbortsStartup(t *testing.T) {
+	t.Parallel()
+
+	key := keyImageRequestsPerMinute
+
+	runAbortCases(t, []abortCase{
+		{
+			name: "negative",
+			yaml: signingKeyLine + key + ": -1\n",
+			wantErrSubstrings: []string{
+				key, testImageRequestsPerMinuteVar, "value -1",
+			},
+		},
+		{
+			name: "not a number",
+			yaml: signingKeyLine + key + ": plenty\n",
+			wantErrSubstrings: []string{
+				key, testImageRequestsPerMinuteVar, "plenty",
+			},
+		},
+		{
+			name:              "fractional",
+			yaml:              signingKeyLine + key + ": 2.5\n",
+			wantErrSubstrings: []string{key, "2.5"},
+		},
+		{
+			name:              "explicit null",
+			yaml:              signingKeyLine + key + ": null\n",
+			wantErrSubstrings: []string{key, nullValueText},
+		},
+	})
+}
+
+// TestImageRequestsPerMinuteFromEnvironment checks that the variable sets the
+// limit over the config file, and that an invalid value in it aborts startup
+// naming the variable and the value.
+func TestImageRequestsPerMinuteFromEnvironment(t *testing.T) {
+	t.Setenv(testImageRequestsPerMinuteVar, "0")
+
+	c, err := configFromYAML(t, signingKeyLine+"image_requests_per_minute: 30\n")
+	if err != nil {
+		t.Fatalf("limit from the environment should load: %v", err)
+	}
+
+	if c.ImageRequestsPerMinute != 0 {
+		t.Errorf("ImageRequestsPerMinute = %d, want 0 from the environment",
+			c.ImageRequestsPerMinute)
+	}
+
+	for _, value := range []string{"-5", "ten"} {
+		t.Run(value, func(t *testing.T) {
+			t.Setenv(testImageRequestsPerMinuteVar, value)
+
+			_, err := configFromYAML(t, signingKeyLine)
+			wantStartupError(t, err, testImageRequestsPerMinuteVar, value)
+		})
+	}
+}

internal/handlers/handlers.go

@@ -4,6 +4,7 @@ package handlers
 import (
 	"context"
 	"encoding/json"
+	"errors"
 	"log/slog"
 	"net/http"
 	"time"
@@ -71,16 +72,21 @@ func New(lc fx.Lifecycle, params Params) (*Handlers, error) {
 	}
 
 	lc.Append(fx.Hook{
-		//nolint:contextcheck // the eviction loop outlives OnStart; OnStop cancels it
+		//nolint:contextcheck // the cache's goroutines outlive OnStart; OnStop stops them
 		OnStart: func(_ context.Context) error {
 			return s.initImageService()
 		},
+		// The pending counts are written here, before the database's stop
+		// hook closes it.
 		OnStop: func(ctx context.Context) error {
 			if s.imgCache == nil {
 				return nil
 			}
 
-			return s.imgCache.StopEviction(ctx)
+			return errors.Join(
+				s.imgCache.StopEviction(ctx),
+				s.imgCache.StopPendingCountWrites(ctx),
+			)
 		},
 	})
 
@@ -100,7 +106,6 @@ func (s *Handlers) WaitForProcessing(ctx context.Context) int {
 func newCacheConfig(cfg *config.Config, log *slog.Logger) imgcache.CacheConfig {
 	return imgcache.CacheConfig{
 		StateDir:           cfg.StateDir,
-		CacheTTL:           imgcache.DefaultCacheTTL,
 		NegativeTTL:        imgcache.DefaultNegativeTTL,
 		MaxBytes:           cfg.CacheMaxBytes,
 		UseDefaultMaxBytes: !cfg.CacheMaxBytesExplicit,
@@ -122,6 +127,9 @@ func (s *Handlers) initImageService() error {
 	// write-pressure passes. No-op when the disk cache is disabled.
 	cache.StartEviction(imgcache.DefaultEvictionInterval)
 
+	// Writes the counts requests could not write by their deadline
+	cache.StartPendingCountWrites()
+
 	// Create the fetcher config
 	fetcherCfg := httpfetcher.DefaultConfig()
 	fetcherCfg.AllowHTTP = s.config.AllowHTTP

internal/handlers/handlers_internal_test.go

@@ -47,7 +47,6 @@ func setupTestHandler(t *testing.T) *testFixtures {
 
 	cache, err := imgcache.NewCache(db, imgcache.CacheConfig{
 		StateDir:    tmpDir,
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {

internal/handlers/image.go

@@ -206,6 +206,16 @@ func parseExpires(query url.Values) (time.Time, error) {
 	return time.Unix(exp, 0), nil
 }
 
+// HandleTooManyRequests answers 429 Too Many Requests with the JSON error
+// body the image routes send for their other errors. The image routes' rate
+// limit calls it for a request over the limit.
+func (s *Handlers) HandleTooManyRequests() http.HandlerFunc {
+	return func(w http.ResponseWriter, _ *http.Request) {
+		s.respondError(w, "too many requests, try again later",
+			http.StatusTooManyRequests)
+	}
+}
+
 // respondImageError maps image retrieval errors to HTTP responses.
 func (s *Handlers) respondImageError(
 	w http.ResponseWriter, req *imgcache.ImageRequest, err error,
@@ -237,6 +247,22 @@ func (s *Handlers) respondImageError(
 		return
 	}
 
+	if errors.Is(err, httpfetcher.ErrInvalidContentType) ||
+		errors.Is(err, imageprocessor.ErrUnsupportedInputFormat) {
+		s.respondError(w, "unsupported source format",
+			http.StatusUnsupportedMediaType)
+
+		return
+	}
+
+	if errors.Is(err, imageprocessor.ErrOutputTooLarge) {
+		s.respondError(w, fmt.Sprintf("size too large for the source: the side "+
+			"computed from its proportions is above %d",
+			imageprocessor.MaxOutputDimension), http.StatusBadRequest)
+
+		return
+	}
+
 	s.respondError(w, "internal error", http.StatusInternalServerError)
 }
 

internal/handlers/image_cache_control_internal_test.go

@@ -31,7 +31,6 @@ func newSignedHostServer(
 
 	cache, err := imgcache.NewCache(setupTestDB(t), imgcache.CacheConfig{
 		StateDir:    t.TempDir(),
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {

internal/handlers/image_errors_internal_test.go

@@ -29,7 +29,6 @@ func newImageRoute(t *testing.T, fetcher httpfetcher.Fetcher) http.Handler {
 
 	cache, err := imgcache.NewCache(setupTestDB(t), imgcache.CacheConfig{
 		StateDir:    t.TempDir(),
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {

internal/handlers/imageenc.go

@@ -161,6 +161,14 @@ func (s *Handlers) handleImageError(w http.ResponseWriter, err error) {
 		errors.Is(err, imageprocessor.ErrTooManyImages):
 		s.respondError(w, "server busy, try again later",
 			http.StatusServiceUnavailable)
+	case errors.Is(err, httpfetcher.ErrInvalidContentType),
+		errors.Is(err, imageprocessor.ErrUnsupportedInputFormat):
+		s.respondError(w, "unsupported source format",
+			http.StatusUnsupportedMediaType)
+	case errors.Is(err, imageprocessor.ErrOutputTooLarge):
+		s.respondError(w, fmt.Sprintf("size too large for the source: the side "+
+			"computed from its proportions is above %d",
+			imageprocessor.MaxOutputDimension), http.StatusBadRequest)
 	default:
 		s.log.Error("image request failed", "error", err)
 		s.respondError(w, "internal error", http.StatusInternalServerError)

internal/handlers/referer_blocklist_internal_test.go

@@ -49,7 +49,6 @@ func newRefererRoutes(t *testing.T) (http.Handler, *Handlers, *countingFetcher)
 
 	cache, err := imgcache.NewCache(setupTestDB(t), imgcache.CacheConfig{
 		StateDir:    t.TempDir(),
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {

internal/handlers/server_busy_internal_test.go

@@ -1,6 +1,7 @@
 package handlers
 
 import (
+	"context"
 	"fmt"
 	"log/slog"
 	"net/http"
@@ -47,3 +48,39 @@ func TestServerBusyAnswers503(t *testing.T) {
 		}
 	}
 }
+
+// TestProcessingTimedOutAnswers503 checks that both image routes answer 503
+// with the server busy error when the image service reports that processing
+// stopped at the deadline of the request doing it, wrapped as the service
+// wraps it. A request that waited for that processing gets this error while
+// its own deadline may have time left.
+func TestProcessingTimedOutAnswers503(t *testing.T) {
+	t.Parallel()
+
+	h := &Handlers{log: slog.New(slog.DiscardHandler)}
+	req := &imgcache.ImageRequest{SourceHost: "img.example.com", SourcePath: "/a.jpg"}
+	err := fmt.Errorf("%w: failed to encode: %w",
+		imgcache.ErrProcessingTimedOut, context.DeadlineExceeded)
+
+	plain := httptest.NewRecorder()
+	h.respondImageError(plain, req, err)
+
+	encrypted := httptest.NewRecorder()
+	h.handleImageError(encrypted, err)
+
+	for route, rec := range map[string]*httptest.ResponseRecorder{
+		"/v1/image/": plain, "/v1/e/": encrypted,
+	} {
+		t.Logf("%s: status = %d, body = %q", route, rec.Code, rec.Body.String())
+
+		if rec.Code != http.StatusServiceUnavailable {
+			t.Errorf("%s: status = %d, want %d",
+				route, rec.Code, http.StatusServiceUnavailable)
+		}
+
+		if !strings.Contains(rec.Body.String(), "server busy, try again later") {
+			t.Errorf("%s: body = %q, want the server busy error",
+				route, rec.Body.String())
+		}
+	}
+}

internal/handlers/unsupported_source_internal_test.go

@@ -0,0 +1,54 @@
+package handlers
+
+import (
+	"fmt"
+	"log/slog"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	"sneak.berlin/go/pixa/internal/httpfetcher"
+	"sneak.berlin/go/pixa/internal/imageprocessor"
+	"sneak.berlin/go/pixa/internal/imgcache"
+)
+
+// TestUnsupportedSourceFormatAnswers415 checks that both image routes answer
+// 415 with a clear error when the upstream answers with a content type pixa
+// does not accept, such as SVG, and when orig is asked of a source in a
+// format the processor cannot name, wrapped as the service wraps them.
+func TestUnsupportedSourceFormatAnswers415(t *testing.T) {
+	t.Parallel()
+
+	h := &Handlers{log: slog.New(slog.DiscardHandler)}
+	req := &imgcache.ImageRequest{SourceHost: "img.example.com", SourcePath: "/a.svg"}
+
+	for _, err := range []error{
+		fmt.Errorf("upstream fetch failed: %w: image/svg+xml",
+			httpfetcher.ErrInvalidContentType),
+		fmt.Errorf("image processing failed: %w: unknown",
+			imageprocessor.ErrUnsupportedInputFormat),
+	} {
+		plain := httptest.NewRecorder()
+		h.respondImageError(plain, req, err)
+
+		encrypted := httptest.NewRecorder()
+		h.handleImageError(encrypted, err)
+
+		for route, rec := range map[string]*httptest.ResponseRecorder{
+			"HandleImage": plain, "HandleImageEnc": encrypted,
+		} {
+			t.Logf("%s for %v: %d %s", route, err, rec.Code, rec.Body.String())
+
+			if rec.Code != http.StatusUnsupportedMediaType {
+				t.Errorf("%s for %v: status = %d, want %d",
+					route, err, rec.Code, http.StatusUnsupportedMediaType)
+			}
+
+			if !strings.Contains(rec.Body.String(), "unsupported source format") {
+				t.Errorf("%s for %v: body = %q, want the unsupported source format error",
+					route, err, rec.Body.String())
+			}
+		}
+	}
+}

internal/httpfetcher/fetch_internal_test.go

@@ -9,6 +9,7 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"slices"
+	"strconv"
 	"strings"
 	"sync"
 	"testing"
@@ -297,6 +298,139 @@ func TestFetchMaxResponseSizeEnforced(t *testing.T) {
 	}
 }
 
+// fetchWithMaxResponseSize serves handler as the fake upstream and fetches
+// its /image route with MaxResponseSize set to maxResponseSize.
+func fetchWithMaxResponseSize(
+	t *testing.T,
+	maxResponseSize int64,
+	handler http.HandlerFunc,
+) *FetchResult {
+	t.Helper()
+
+	srv := httptest.NewServer(handler)
+	t.Cleanup(srv.Close)
+
+	cfg := DefaultConfig()
+	cfg.MaxResponseSize = maxResponseSize
+
+	f, _ := newServerFetcher(t, srv, cfg)
+
+	res := fetchImage(t, f, "/image")
+	t.Cleanup(func() { _ = res.Content.Close() })
+
+	return res
+}
+
+// TestFetchContentLengthBodyAtMaxResponseSize checks that a body sent with a
+// Content-Length is read whole at exactly MaxResponseSize bytes and refused
+// at one byte more.
+func TestFetchContentLengthBodyAtMaxResponseSize(t *testing.T) {
+	t.Parallel()
+
+	const maxResponseSize = 8
+
+	for _, tc := range []struct {
+		size    int
+		wantErr error
+	}{
+		{maxResponseSize, nil},
+		{maxResponseSize + 1, ErrResponseTooLarge},
+	} {
+		t.Run(strconv.Itoa(tc.size)+" bytes", func(t *testing.T) {
+			t.Parallel()
+
+			body := strings.Repeat("x", tc.size)
+
+			res := fetchWithMaxResponseSize(t, maxResponseSize,
+				func(w http.ResponseWriter, _ *http.Request) {
+					w.Header().Set("Content-Type", contentTypeJPEG)
+					w.Header().Set("Content-Length", strconv.Itoa(tc.size))
+					_, _ = io.WriteString(w, body)
+				})
+
+			if res.ContentLength != int64(tc.size) {
+				t.Fatalf("ContentLength = %d, want %d", res.ContentLength, tc.size)
+			}
+
+			data, err := io.ReadAll(res.Content)
+			t.Logf("body of %d bytes: read %d, error %v", tc.size, len(data), err)
+
+			if !errors.Is(err, tc.wantErr) {
+				t.Fatalf("read error = %v, want %v", err, tc.wantErr)
+			}
+
+			if err == nil && string(data) != body {
+				t.Errorf("read %q, want %q", data, body)
+			}
+		})
+	}
+}
+
+// TestFetchChunkedBodyAtMaxResponseSize checks that a chunked body is read
+// whole at exactly MaxResponseSize bytes and refused at one byte more. The
+// upstream sends the end of the body only once the test has read
+// MaxResponseSize bytes, so the read after them comes before the end of the
+// body has arrived, as it can from a real upstream.
+func TestFetchChunkedBodyAtMaxResponseSize(t *testing.T) {
+	t.Parallel()
+
+	const maxResponseSize = 8
+
+	for _, tc := range []struct {
+		size    int
+		wantErr error
+	}{
+		{maxResponseSize, nil},
+		{maxResponseSize + 1, ErrResponseTooLarge},
+	} {
+		t.Run(strconv.Itoa(tc.size)+" bytes", func(t *testing.T) {
+			t.Parallel()
+
+			body := strings.Repeat("x", tc.size)
+			endBody := make(chan struct{})
+
+			res := fetchWithMaxResponseSize(t, maxResponseSize,
+				func(w http.ResponseWriter, r *http.Request) {
+					w.Header().Set("Content-Type", contentTypeJPEG)
+					_, _ = io.WriteString(w, body)
+					// With no Content-Length, flushing sends the body
+					// chunked; its end is sent when the handler returns.
+					_ = http.NewResponseController(w).Flush()
+
+					select {
+					case <-endBody:
+					case <-r.Context().Done():
+					}
+				})
+
+			if res.ContentLength != -1 {
+				t.Fatalf("ContentLength = %d, want -1 (chunked)", res.ContentLength)
+			}
+
+			head := make([]byte, maxResponseSize)
+
+			_, err := io.ReadFull(res.Content, head)
+			if err != nil {
+				t.Fatalf("read the first %d bytes: %v", maxResponseSize, err)
+			}
+
+			close(endBody)
+
+			rest, err := io.ReadAll(res.Content)
+			data := string(head) + string(rest)
+			t.Logf("body of %d bytes: read %d, error %v", tc.size, len(data), err)
+
+			if !errors.Is(err, tc.wantErr) {
+				t.Fatalf("read error = %v, want %v", err, tc.wantErr)
+			}
+
+			if err == nil && data != body {
+				t.Errorf("read %q, want %q", data, body)
+			}
+		})
+	}
+}
+
 func TestFetchSemaphoreReleasedOnError(t *testing.T) {
 	t.Parallel()
 

internal/httpfetcher/httpfetcher.go

@@ -158,7 +158,6 @@ func DefaultConfig() *Config {
 			contentTypeWebP,
 			contentTypeAVIF,
 			contentTypeJXL,
-			contentTypeSVG,
 		},
 		AllowHTTP:             false,
 		MaxConnectionsPerHost: DefaultMaxConnectionsPerHost,
@@ -661,14 +660,33 @@ func dialSSRFSafe(
 	return nil, fmt.Errorf("%w to %s", errConnectFailed, host)
 }
 
-// limitedReader wraps a reader and limits the number of bytes read.
+// limitedReader wraps a reader and refuses a body longer than remaining
+// bytes with ErrResponseTooLarge. remaining goes below 0 once the body is
+// found to be longer.
 type limitedReader struct {
 	reader    io.Reader
 	remaining int64
 }
 
+// Read reads at most the remaining bytes. Once they are read, it reads one
+// more byte to tell the end of the body from more data: at the end it
+// returns the wrapped reader's error, io.EOF; with more data it returns
+// ErrResponseTooLarge, as does every later Read.
 func (r *limitedReader) Read(p []byte) (int, error) {
-	if r.remaining <= 0 {
+	if r.remaining < 0 {
+		return 0, ErrResponseTooLarge
+	}
+
+	if r.remaining == 0 {
+		var extra [1]byte
+
+		n, err := r.reader.Read(extra[:])
+		r.remaining -= int64(n)
+
+		if n == 0 {
+			return 0, err
+		}
+
 		return 0, ErrResponseTooLarge
 	}
 

internal/imageprocessor/computed_side_internal_test.go

@@ -0,0 +1,52 @@
+package imageprocessor
+
+import (
+	"bytes"
+	"errors"
+	"testing"
+)
+
+// TestImageProcessor_RefusesComputedSideAboveMaximum verifies that a size
+// with one side 0, which Process computes from the input's proportions, is
+// served when that side comes out at MaxOutputDimension and refused with
+// ErrOutputTooLarge when it would be above. A 4096x1 or 1x4096 input given
+// 2 for its short side has 8192 for its long side.
+func TestImageProcessor_RefusesComputedSideAboveMaximum(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name                    string
+		inputWidth, inputHeight int
+		size                    Size
+		wantErr                 error
+	}{
+		{"width at the maximum", 4096, 1, Size{Height: 2}, nil},
+		{"width above the maximum", 4096, 1, Size{Height: 3}, ErrOutputTooLarge},
+		{"height at the maximum", 1, 4096, Size{Width: 2}, nil},
+		{"height above the maximum", 1, 4096, Size{Width: 3}, ErrOutputTooLarge},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			input := createTestJPEG(t, tt.inputWidth, tt.inputHeight)
+
+			result, err := New(Params{}).Process(t.Context(),
+				bytes.NewReader(input),
+				&Request{Size: tt.size, Format: FormatJPEG, FitMode: FitCover})
+			if result != nil {
+				t.Logf("Process() made %dx%d", result.Width, result.Height)
+
+				if max(result.Width, result.Height) != MaxOutputDimension {
+					t.Errorf("Process() made %dx%d, want a side of %d",
+						result.Width, result.Height, MaxOutputDimension)
+				}
+			}
+
+			if !errors.Is(err, tt.wantErr) {
+				t.Errorf("Process() error = %v, want %v", err, tt.wantErr)
+			}
+		})
+	}
+}

internal/imageprocessor/deadline_internal_test.go

@@ -0,0 +1,118 @@
+package imageprocessor
+
+import (
+	"bytes"
+	"context"
+	"errors"
+	"io"
+	"testing"
+	"time"
+)
+
+// readerAfterDeadline is a Process input whose first Read waits until done
+// is closed, so the deadline passes after Process has taken its slot and
+// before it encodes.
+type readerAfterDeadline struct {
+	data   *bytes.Reader
+	done   <-chan struct{}
+	waited bool
+}
+
+func (r *readerAfterDeadline) Read(p []byte) (int, error) {
+	if !r.waited {
+		r.waited = true
+
+		<-r.done
+	}
+
+	return r.data.Read(p)
+}
+
+// TestProcessStopsWhenDeadlinePassed checks that Process returns
+// context.DeadlineExceeded, and holds no slot afterwards, when its context's
+// deadline has passed before it starts or while it reads its input. libvips
+// cannot stop an encode once it has begun, so Process must not begin one for
+// a request that has ended.
+func TestProcessStopsWhenDeadlinePassed(t *testing.T) {
+	t.Parallel()
+
+	input := createTestJPEG(t, 50, 50)
+
+	t.Run("before Process starts", func(t *testing.T) {
+		t.Parallel()
+
+		ctx, cancel := context.WithDeadline(t.Context(),
+			time.Now().Add(-time.Second))
+		defer cancel()
+
+		checkProcessStopsAtDeadline(ctx, t, bytes.NewReader(input))
+	})
+
+	t.Run("while Process reads its input", func(t *testing.T) {
+		t.Parallel()
+
+		ctx, cancel := context.WithTimeout(t.Context(), 100*time.Millisecond)
+		defer cancel()
+
+		checkProcessStopsAtDeadline(ctx, t, &readerAfterDeadline{
+			data: bytes.NewReader(input), done: ctx.Done(),
+		})
+	})
+}
+
+// checkProcessStopsAtDeadline runs Process on input with ctx, whose deadline
+// has passed by the time Process could encode, and fails the test unless
+// Process returns context.DeadlineExceeded and gives its slot back.
+func checkProcessStopsAtDeadline(
+	ctx context.Context, t *testing.T, input io.Reader,
+) {
+	t.Helper()
+
+	proc := New(Params{MaxConcurrentProcessing: 1})
+
+	result, err := proc.Process(ctx, input, smallJPEGRequest())
+	if result != nil {
+		t.Logf("Process() made a %dx%d image after its deadline passed",
+			result.Width, result.Height)
+
+		_ = result.Content.Close()
+	}
+
+	t.Logf("Process() error = %v", err)
+
+	if !errors.Is(err, context.DeadlineExceeded) {
+		t.Errorf("Process() error = %v, want context.DeadlineExceeded", err)
+	}
+
+	if held := len(proc.processingSemaphore); held != 0 {
+		t.Errorf("%d slots still held after Process() returned", held)
+	}
+}
+
+// inputNotToBeRead is a Process input that fails the test when it is read.
+type inputNotToBeRead struct {
+	t *testing.T
+}
+
+func (r inputNotToBeRead) Read([]byte) (int, error) {
+	r.t.Error("Process read its input although its deadline had passed")
+
+	return 0, io.EOF
+}
+
+// TestProcessReadsNoInputOnceDeadlinePassed checks that Process, called with
+// a context whose deadline passed before it starts, gives its slot back
+// without reading its input. With a slot free, the select in acquireSlot
+// takes the slot or sees the ended context at random, so Process is called
+// many times.
+func TestProcessReadsNoInputOnceDeadlinePassed(t *testing.T) {
+	t.Parallel()
+
+	ctx, cancel := context.WithDeadline(t.Context(),
+		time.Now().Add(-time.Second))
+	defer cancel()
+
+	for range 100 {
+		checkProcessStopsAtDeadline(ctx, t, inputNotToBeRead{t: t})
+	}
+}

internal/imageprocessor/fit_outside_internal_test.go

@@ -0,0 +1,62 @@
+package imageprocessor
+
+import (
+	"bytes"
+	"errors"
+	"testing"
+)
+
+// TestImageProcessor_RefusesFitOutsideAboveMaximum verifies that FitOutside,
+// which scales the input to cover the requested size and so past it on one
+// side, is refused with ErrOutputTooLarge when that side would come out above
+// MaxOutputDimension, and otherwise makes the scaled size. A 1024x256 or
+// 256x1024 input scaled by 8 has 8192 for its long side.
+func TestImageProcessor_RefusesFitOutsideAboveMaximum(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name                    string
+		inputWidth, inputHeight int
+		size                    Size
+		// wantWidth and wantHeight are the output's, 0 for a refusal
+		wantWidth, wantHeight int
+	}{
+		{"width above the maximum", 3000, 600, Size{Width: 1, Height: 8192}, 0, 0},
+		{"height above the maximum", 600, 3000, Size{Width: 8192, Height: 1}, 0, 0},
+		{"width at the maximum", 1024, 256, Size{Width: 1, Height: 2048}, 8192, 2048},
+		{"height at the maximum", 256, 1024, Size{Width: 2048, Height: 1}, 2048, 8192},
+		{"within the maximum", 3000, 600, Size{Width: 1, Height: 1200}, 6000, 1200},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			input := createTestJPEG(t, tt.inputWidth, tt.inputHeight)
+
+			result, err := New(Params{}).Process(t.Context(),
+				bytes.NewReader(input),
+				&Request{Size: tt.size, Format: FormatJPEG, FitMode: FitOutside})
+			if result != nil {
+				t.Logf("Process() made %dx%d", result.Width, result.Height)
+			}
+
+			if tt.wantWidth == 0 {
+				if !errors.Is(err, ErrOutputTooLarge) {
+					t.Errorf("Process() error = %v, want %v", err, ErrOutputTooLarge)
+				}
+
+				return
+			}
+
+			if err != nil {
+				t.Fatalf("Process() error = %v", err)
+			}
+
+			if result.Width != tt.wantWidth || result.Height != tt.wantHeight {
+				t.Errorf("Process() made %dx%d, want %dx%d",
+					result.Width, result.Height, tt.wantWidth, tt.wantHeight)
+			}
+		})
+	}
+}

internal/imageprocessor/imageprocessor.go

@@ -133,10 +133,26 @@ var ErrInputTooLarge = errors.New("input image dimensions exceed maximum")
 // configured byte limit.
 var ErrInputDataTooLarge = errors.New("input data exceeds maximum allowed size")
 
+// MaxOutputDimension is the maximum width or height of an output image, and
+// so of a requested size (imgcache.MaxDimension).
+const MaxOutputDimension = 8192
+
+// ErrOutputTooLarge is returned, before resizing, when the output would have a
+// side above MaxOutputDimension: the side of a size with one side 0, computed
+// from the input's proportions, or with FitOutside, the side scaled past the
+// requested size.
+var ErrOutputTooLarge = errors.New("output image dimensions exceed maximum")
+
 // ErrUnsupportedOutputFormat is returned when the requested output format is
 // not supported.
 var ErrUnsupportedOutputFormat = errors.New("unsupported output format")
 
+// ErrUnsupportedInputFormat is returned, before the input is decoded, when
+// libvips reads the input's first bytes as another format than the source
+// formats, or as none: for example an SVG whose first bytes pass the
+// first-bytes check for AVIF.
+var ErrUnsupportedInputFormat = errors.New("unsupported input format")
+
 // ErrTooManyImages is returned when MaxConcurrentProcessing images are being
 // processed and none finishes within ProcessingWaitTimeout.
 var ErrTooManyImages = errors.New("too many images being processed at once")
@@ -192,7 +208,9 @@ func New(params Params) *ImageProcessor {
 // Process transforms an image according to the request. When
 // MaxConcurrentProcessing images are already being processed, it waits up
 // to ProcessingWaitTimeout for one to finish, then fails with
-// ErrTooManyImages.
+// ErrTooManyImages. When ctx has ended once Process has a slot, or before it
+// encodes, it returns ctx's error and gives the slot back. libvips cannot stop
+// an encode once it has begun, so an encode keeps its slot until it finishes.
 func (p *ImageProcessor) Process(
 	ctx context.Context,
 	input io.Reader,
@@ -219,6 +237,17 @@ func (p *ImageProcessor) Process(
 		return nil, ErrInputDataTooLarge
 	}
 
+	// govips picks the libvips loader from the input's first bytes with
+	// DetermineImageType. Input whose first bytes it reads as no source
+	// format, such as SVG, which librsvg would render, is refused before
+	// libvips reads it.
+	inputFormat := p.detectFormat(vips.DetermineImageType(data))
+
+	sourceFormat, err := p.formatFromString(inputFormat)
+	if err != nil {
+		return nil, err
+	}
+
 	// Decode image
 	img, err := vips.NewImageFromBuffer(data)
 	if err != nil {
@@ -237,16 +266,16 @@ func (p *ImageProcessor) Process(
 	origWidth := img.Width()
 	origHeight := img.Height()
 
-	// Detect input format
-	inputFormat := p.detectFormat(img)
-
 	// Validate input dimensions to prevent DoS via decompression bombs
 	if origWidth > MaxInputDimension || origHeight > MaxInputDimension {
 		return nil, ErrInputTooLarge
 	}
 
-	// Determine target dimensions
-	targetWidth, targetHeight := targetDimensions(req.Size, origWidth, origHeight)
+	// Determine target dimensions; too large ones are refused before resizing
+	targetWidth, targetHeight, err := targetDimensions(req.Size, origWidth, origHeight)
+	if err != nil {
+		return nil, err
+	}
 
 	// Resize if needed
 	if targetWidth != origWidth || targetHeight != origHeight {
@@ -259,11 +288,11 @@ func (p *ImageProcessor) Process(
 	// orig is the source's own format; encode refuses an empty format
 	outputFormat := req.Format
 	if outputFormat == FormatOriginal {
-		outputFormat = p.formatFromString(inputFormat)
+		outputFormat = sourceFormat
 	}
 
 	// Encode to target format
-	output, err := p.encode(img, outputFormat, req.Quality)
+	output, err := p.encode(ctx, img, outputFormat, req.Quality)
 	if err != nil {
 		return nil, fmt.Errorf("failed to encode: %w", err)
 	}
@@ -282,21 +311,29 @@ func (p *ImageProcessor) Process(
 
 // targetDimensions calculates the output dimensions for a requested size,
 // scaling proportionally when only one dimension is given and keeping the
-// original dimensions when both are zero.
-func targetDimensions(size Size, origWidth, origHeight int) (int, int) {
+// original dimensions when both are zero. A side scaled proportionally can
+// come out above MaxOutputDimension, which is ErrOutputTooLarge.
+func targetDimensions(size Size, origWidth, origHeight int) (int, int, error) {
+	width, height := size.Width, size.Height
+
 	switch {
-	case size.Width == 0 && size.Height == 0:
+	case width == 0 && height == 0:
 		// Both are 0: keep original size
-		return origWidth, origHeight
-	case size.Width == 0:
+		return origWidth, origHeight, nil
+	case width == 0:
 		// Only height specified: calculate width proportionally
-		return origWidth * size.Height / origHeight, size.Height
-	case size.Height == 0:
+		width = origWidth * height / origHeight
+	case height == 0:
 		// Only width specified: calculate height proportionally
-		return size.Width, origHeight * size.Width / origWidth
-	default:
-		return size.Width, size.Height
+		height = origHeight * width / origWidth
 	}
+
+	if width > MaxOutputDimension || height > MaxOutputDimension {
+		return 0, 0, fmt.Errorf("%w: %dx%d is above %d", ErrOutputTooLarge,
+			width, height, MaxOutputDimension)
+	}
+
+	return width, height, nil
 }
 
 // MIME types for the supported image formats.
@@ -382,31 +419,33 @@ func (p *ImageProcessor) WaitForProcessing(ctx context.Context) int {
 
 // acquireSlot takes a slot in processingSemaphore, waiting at most
 // processingWaitTimeout for one to free up, and returns the func that gives
-// it back. A free slot is taken even when ctx has ended; only the wait for
-// one stops when ctx ends, as the rest of Process does not check ctx.
+// it back. When ctx has ended by the time it has a slot, it gives the slot
+// back and returns ctx's error.
 func (p *ImageProcessor) acquireSlot(ctx context.Context) (func(), error) {
-	release := func() { <-p.processingSemaphore }
-
 	select {
 	case p.processingSemaphore <- struct{}{}:
-		return release, nil
-	default:
-	}
-
-	select {
-	case p.processingSemaphore <- struct{}{}:
-		return release, nil
 	case <-time.After(p.processingWaitTimeout):
 		return nil, ErrTooManyImages
 	case <-ctx.Done():
 		return nil, ctx.Err()
 	}
+
+	release := func() { <-p.processingSemaphore }
+
+	// select takes the slot even when ctx has ended at the same moment
+	err := ctx.Err()
+	if err != nil {
+		release()
+
+		return nil, err
+	}
+
+	return release, nil
 }
 
-// detectFormat returns the format string from a vips image.
-func (p *ImageProcessor) detectFormat(img *vips.ImageRef) string {
-	format := img.Format()
-	switch format {
+// detectFormat returns the format string for a govips image type.
+func (p *ImageProcessor) detectFormat(imageType vips.ImageType) string {
+	switch imageType {
 	case vips.ImageTypeJPEG:
 		return "jpeg"
 	case vips.ImageTypePNG:
@@ -428,7 +467,9 @@ func (p *ImageProcessor) detectFormat(img *vips.ImageRef) string {
 	}
 }
 
-// resize resizes the image according to the fit mode.
+// resize resizes the image according to the fit mode. With FitOutside one
+// side can come out larger than requested; when it would be above
+// MaxOutputDimension, resize returns ErrOutputTooLarge without resizing.
 func (p *ImageProcessor) resize(
 	img *vips.ImageRef, width, height int, fit FitMode,
 ) error {
@@ -476,6 +517,12 @@ func (p *ImageProcessor) resize(
 		newW := int(float64(imgW) * scale)
 		newH := int(float64(imgH) * scale)
 
+		// The dimension that exceeds the target is held to the maximum too
+		if newW > MaxOutputDimension || newH > MaxOutputDimension {
+			return fmt.Errorf("%w: %dx%d is above %d", ErrOutputTooLarge,
+				newW, newH, MaxOutputDimension)
+		}
+
 		return img.Thumbnail(newW, newH, vips.InterestingNone)
 
 	default:
@@ -485,10 +532,17 @@ func (p *ImageProcessor) resize(
 
 const defaultQuality = 85
 
-// encode encodes an image to the specified format.
+// encode encodes an image to the specified format. libvips cannot stop an
+// encode once it has begun, so when ctx has ended encode returns ctx's error
+// without beginning one.
 func (p *ImageProcessor) encode(
-	img *vips.ImageRef, format Format, quality int,
+	ctx context.Context, img *vips.ImageRef, format Format, quality int,
 ) ([]byte, error) {
+	err := ctx.Err()
+	if err != nil {
+		return nil, err
+	}
+
 	if quality <= 0 {
 		quality = defaultQuality
 	}
@@ -498,7 +552,7 @@ func (p *ImageProcessor) encode(
 	// libvips' built-in profile; govips' own sRGB path variable is set on
 	// first use but read without a lock, so concurrent requests race on it.
 	if img.HasICCProfile() {
-		err := img.TransformICCProfileWithFallback("srgb", "srgb")
+		err = img.TransformICCProfileWithFallback("srgb", "srgb")
 		if err != nil {
 			return nil, fmt.Errorf("failed to convert to sRGB: %w", err)
 		}
@@ -687,22 +741,23 @@ func jxlDistance(quality int) float64 {
 	return 53.0/3000.0*q*q - 23.0/20.0*q + 25.0
 }
 
-// formatFromString converts a format string to Format.
-func (p *ImageProcessor) formatFromString(format string) Format {
+// formatFromString converts a format string from detectFormat to Format. A
+// format detectFormat could not name is ErrUnsupportedInputFormat.
+func (p *ImageProcessor) formatFromString(format string) (Format, error) {
 	switch format {
 	case "jpeg":
-		return FormatJPEG
+		return FormatJPEG, nil
 	case "png":
-		return FormatPNG
+		return FormatPNG, nil
 	case "gif":
-		return FormatGIF
+		return FormatGIF, nil
 	case "webp":
-		return FormatWebP
+		return FormatWebP, nil
 	case string(FormatAVIF):
-		return FormatAVIF
+		return FormatAVIF, nil
 	case string(FormatJXL):
-		return FormatJXL
+		return FormatJXL, nil
 	default:
-		return FormatJPEG
+		return "", fmt.Errorf("%w: %s", ErrUnsupportedInputFormat, format)
 	}
 }

internal/imageprocessor/orig_format_internal_test.go

@@ -0,0 +1,39 @@
+package imageprocessor
+
+import (
+	"bytes"
+	"errors"
+	"testing"
+
+	"github.com/davidbyttow/govips/v2/vips"
+)
+
+// TestImageProcessor_OrigOfUnnamedFormatRefused verifies that orig of a TIFF
+// input, which is not a source format, is refused with
+// ErrUnsupportedInputFormat before it is decoded, and not saved as JPEG.
+func TestImageProcessor_OrigOfUnnamedFormatRefused(t *testing.T) {
+	t.Parallel()
+
+	img, err := vips.NewImageFromBuffer(createTestPNG(t, 20, 20))
+	if err != nil {
+		t.Fatalf("failed to decode test PNG: %v", err)
+	}
+
+	t.Cleanup(img.Close)
+
+	tiff, _, err := img.ExportTiff(vips.NewTiffExportParams())
+	if err != nil {
+		t.Fatalf("failed to save test TIFF: %v", err)
+	}
+
+	result, err := New(Params{}).Process(
+		t.Context(), bytes.NewReader(tiff), &Request{Format: FormatOriginal},
+	)
+	if result != nil {
+		t.Logf("Process() returned %s", result.ContentType)
+	}
+
+	if !errors.Is(err, ErrUnsupportedInputFormat) {
+		t.Errorf("Process() error = %v, want %v", err, ErrUnsupportedInputFormat)
+	}
+}

internal/imageprocessor/source_formats_internal_test.go

@@ -0,0 +1,46 @@
+package imageprocessor
+
+import (
+	"bytes"
+	"io"
+	"testing"
+)
+
+// TestImageProcessor_LoadsEachSourceFormat verifies that an image in each of
+// the six source formats, saved by the processor itself, is loaded and saved
+// again in its own format (orig).
+func TestImageProcessor_LoadsEachSourceFormat(t *testing.T) {
+	t.Parallel()
+
+	proc := New(Params{})
+
+	for _, format := range []Format{
+		FormatJPEG, FormatPNG, FormatGIF, FormatWebP, FormatAVIF, FormatJXL,
+	} {
+		saved, err := proc.Process(t.Context(),
+			bytes.NewReader(createTestPNG(t, 20, 20)), &Request{Format: format})
+		if err != nil {
+			t.Fatalf("saving the test image as %s: %v", format, err)
+		}
+
+		source, err := io.ReadAll(saved.Content)
+		if err != nil {
+			t.Fatalf("reading the %s image: %v", format, err)
+		}
+
+		result, err := proc.Process(t.Context(), bytes.NewReader(source),
+			&Request{Format: FormatOriginal})
+		if err != nil {
+			t.Errorf("orig of the %s image: %v", format, err)
+
+			continue
+		}
+
+		t.Logf("orig of the %s image: %s", format, result.ContentType)
+
+		if result.ContentType != FormatToMIME(format) {
+			t.Errorf("orig of the %s image is %s, want %s",
+				format, result.ContentType, FormatToMIME(format))
+		}
+	}
+}

internal/imgcache/cache.go

@@ -11,6 +11,8 @@ import (
 	"io"
 	"log/slog"
 	"path/filepath"
+	"sync"
+	"sync/atomic"
 	"time"
 
 	lru "github.com/hashicorp/golang-lru/v2"
@@ -33,7 +35,6 @@ const metaCacheSize = 10000
 // CacheConfig holds cache configuration.
 type CacheConfig struct {
 	StateDir    string
-	CacheTTL    time.Duration
 	NegativeTTL time.Duration
 
 	// MaxBytes is the disk cache size limit in bytes that eviction
@@ -79,6 +80,27 @@ type Cache struct {
 	evictionDone     chan struct{}
 	evictionCancel   context.CancelFunc
 
+	// The pending counts: hits, misses, upstream fetches and transforms
+	// whose write to the database missed its request's deadline, kept
+	// here until they are written by the goroutine that
+	// StartPendingCountWrites starts. As for eviction, the channels are
+	// created in NewCache: pendingCountsAdded wakes that goroutine, and
+	// pendingCountsDone is closed when it returns. pendingCountsCancel,
+	// set by StartPendingCountWrites, cancels its context, which tells it
+	// to finish.
+	pendingHits               atomic.Int64
+	pendingMisses             atomic.Int64
+	pendingUpstreamFetches    atomic.Int64
+	pendingUpstreamFetchBytes atomic.Int64
+	pendingTransforms         atomic.Int64
+	pendingCountsAdded        chan struct{}
+	pendingCountsDone         chan struct{}
+	pendingCountsCancel       context.CancelFunc
+
+	// Held through each write of the pending counts, and by
+	// readHitAndMissCounts to read them
+	pendingCountsWriteMutex sync.Mutex
+
 	// metaCache holds the content types of the variants most recently
 	// stored or served, so a hit does not read the variant's .meta file.
 	// It never stands in for the variant file, which is always opened.
@@ -139,6 +161,9 @@ func newCache(
 		metaCache:        metaCache,
 		contentLocks:     newContentLock(),
 
+		pendingCountsAdded: make(chan struct{}, 1),
+		pendingCountsDone:  make(chan struct{}),
+
 		reconciliationPageSize: defaultReconciliationPageSize,
 	}
 
@@ -474,12 +499,7 @@ func (c *Cache) CleanExpired(ctx context.Context) error {
 func (c *Cache) Stats(ctx context.Context) (*CacheStats, error) {
 	var stats CacheStats
 
-	// Fetch hit/miss counts from the stats table
-	err := c.db.QueryRowContext(ctx, `
-		SELECT hit_count, miss_count
-		FROM cache_stats WHERE id = 1
-	`).Scan(&stats.HitCount, &stats.MissCount)
-
+	err := c.readHitAndMissCounts(ctx, &stats)
 	if err != nil && !errors.Is(err, sql.ErrNoRows) {
 		return nil, fmt.Errorf("failed to get cache stats: %w", err)
 	}
@@ -511,19 +531,30 @@ func (c *Cache) Stats(ctx context.Context) (*CacheStats, error) {
 }
 
 // IncrementStats counts a cache hit or miss, and an upstream fetch that read
-// fetchBytes bytes, as IncrementUpstreamFetch does.
+// fetchBytes bytes, as IncrementUpstreamFetch does. Like the other Increment
+// methods, it writes to the database with ctx's deadline but not its
+// cancellation, so an ended request is still counted but does not wait for
+// the database past its deadline; a count not written by then becomes a
+// pending count (see StartPendingCountWrites).
 func (c *Cache) IncrementStats(ctx context.Context, hit bool, fetchBytes int64) {
+	countCtx, cancel := withoutCancelKeepingDeadline(ctx)
+	defer cancel()
+
 	var err error
 
+	pendingCount := &c.pendingMisses
+
 	if hit {
-		_, err = c.db.ExecContext(ctx, `
+		pendingCount = &c.pendingHits
+
+		_, err = c.db.ExecContext(countCtx, `
 			UPDATE cache_stats
 			SET hit_count = hit_count + 1,
 			    last_updated_at = CURRENT_TIMESTAMP
 			WHERE id = 1
 		`)
 	} else {
-		_, err = c.db.ExecContext(ctx, `
+		_, err = c.db.ExecContext(countCtx, `
 			UPDATE cache_stats
 			SET miss_count = miss_count + 1,
 			    last_updated_at = CURRENT_TIMESTAMP
@@ -531,7 +562,10 @@ func (c *Cache) IncrementStats(ctx context.Context, hit bool, fetchBytes int64)
 		`)
 	}
 
-	if err != nil {
+	switch {
+	case errors.Is(err, context.DeadlineExceeded):
+		c.addPendingCount(pendingCount, 1)
+	case err != nil:
 		c.log.Warn("failed to count cache hit or miss", "hit", hit, "error", err)
 	}
 
@@ -545,14 +579,22 @@ func (c *Cache) IncrementUpstreamFetch(ctx context.Context, fetchBytes int64) {
 		return
 	}
 
-	_, err := c.db.ExecContext(ctx, `
+	countCtx, cancel := withoutCancelKeepingDeadline(ctx)
+	defer cancel()
+
+	_, err := c.db.ExecContext(countCtx, `
 		UPDATE cache_stats
 		SET upstream_fetch_count = upstream_fetch_count + 1,
 		    upstream_fetch_bytes = upstream_fetch_bytes + ?,
 		    last_updated_at = CURRENT_TIMESTAMP
 		WHERE id = 1
 	`, fetchBytes)
-	if err != nil {
+
+	switch {
+	case errors.Is(err, context.DeadlineExceeded):
+		c.addPendingCount(&c.pendingUpstreamFetches, 1)
+		c.addPendingCount(&c.pendingUpstreamFetchBytes, fetchBytes)
+	case err != nil:
 		c.log.Warn("failed to count upstream fetch",
 			"fetch_bytes", fetchBytes, "error", err)
 	}
@@ -560,17 +602,43 @@ func (c *Cache) IncrementUpstreamFetch(ctx context.Context, fetchBytes int64) {
 
 // IncrementTransformCount counts one image transcoded by the image processor.
 func (c *Cache) IncrementTransformCount(ctx context.Context) {
-	_, err := c.db.ExecContext(ctx, `
+	countCtx, cancel := withoutCancelKeepingDeadline(ctx)
+	defer cancel()
+
+	_, err := c.db.ExecContext(countCtx, `
 		UPDATE cache_stats
 		SET transform_count = transform_count + 1,
 		    last_updated_at = CURRENT_TIMESTAMP
 		WHERE id = 1
 	`)
-	if err != nil {
+
+	switch {
+	case errors.Is(err, context.DeadlineExceeded):
+		c.addPendingCount(&c.pendingTransforms, 1)
+	case err != nil:
 		c.log.Warn("failed to count transform", "error", err)
 	}
 }
 
+// readHitAndMissCounts sets the hit and miss counts of stats to those of the
+// cache_stats row plus the pending hits and misses (see
+// StartPendingCountWrites). It holds pendingCountsWriteMutex, so that no
+// write of the pending counts falls between the two reads.
+func (c *Cache) readHitAndMissCounts(ctx context.Context, stats *CacheStats) error {
+	c.pendingCountsWriteMutex.Lock()
+	defer c.pendingCountsWriteMutex.Unlock()
+
+	err := c.db.QueryRowContext(ctx, `
+		SELECT hit_count, miss_count
+		FROM cache_stats WHERE id = 1
+	`).Scan(&stats.HitCount, &stats.MissCount)
+
+	stats.HitCount += c.pendingHits.Load()
+	stats.MissCount += c.pendingMisses.Load()
+
+	return err
+}
+
 // loadVariantWithMeta is GetVariant for a variant metaCache does not
 // hold: it reads the content type from the variant's .meta file and
 // keeps it in metaCache, unless a StoreVariant has put one there

internal/imgcache/cache_internal_test.go

@@ -102,7 +102,6 @@ func setupTestCache(t *testing.T) *Cache {
 
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    tmpDir,
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {
@@ -241,7 +240,6 @@ func TestCache_NegativeCacheExpiry(t *testing.T) {
 	// Very short negative TTL for testing
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    tmpDir,
-		CacheTTL:    time.Hour,
 		NegativeTTL: 1 * time.Millisecond,
 	})
 	if err != nil {

internal/imgcache/concurrent_misses_internal_test.go

@@ -15,6 +15,7 @@ import (
 
 	"github.com/getsentry/sentry-go"
 	"sneak.berlin/go/pixa/internal/httpfetcher"
+	"sneak.berlin/go/pixa/internal/imageprocessor"
 	"sneak.berlin/go/pixa/internal/magic"
 )
 
@@ -435,6 +436,47 @@ func TestService_Get_ReturnsByItsDeadline(t *testing.T) {
 	}
 }
 
+// TestService_Get_WaiterGetsProcessingTimedOut has a request with no deadline
+// join the held fetch of a first request with a short one. With the only
+// processing slot held, the first request's transcode waits for it until the
+// first request's deadline ends, and stops then. The waiting request, whose
+// own context has not ended, must get ErrProcessingTimedOut, which the image
+// routes answer 503 as when pixa is busy.
+func TestService_Get_WaiterGetsProcessingTimedOut(t *testing.T) {
+	t.Parallel()
+
+	svc, fixtures, fetcher := setupHeldFetchService(t)
+	svc.processor = imageprocessor.New(
+		imageprocessor.Params{MaxConcurrentProcessing: 1},
+	)
+
+	holdProcessingSlot(t, svc.processor)
+
+	const firstTimeout = 500 * time.Millisecond
+
+	firstCtx, cancelFirst := context.WithTimeout(t.Context(), firstTimeout)
+	defer cancelFirst()
+
+	first := startGet(firstCtx, svc, photoVariant(fixtures, 85, FitCover))
+
+	<-fetcher.started
+
+	waiter := startGet(t.Context(), svc, photoVariant(fixtures, 85, FitCover))
+
+	time.Sleep(arrivalWait)
+	fetcher.releaseFetches()
+
+	t.Logf("first request: Get() error = %v", (<-first).err)
+
+	got := <-waiter
+	t.Logf("waiting request: Get() error = %v", got.err)
+
+	if !errors.Is(got.err, ErrProcessingTimedOut) {
+		t.Errorf("waiting request: Get() error = %v, want %v",
+			got.err, ErrProcessingTimedOut)
+	}
+}
+
 // panickingFetcher panics on every fetch.
 type panickingFetcher struct{}
 

internal/imgcache/eviction_internal_test.go

@@ -70,7 +70,6 @@ func newEvictionTestCache(t *testing.T, maxBytes int64) (*Cache, string) {
 	// backwards compatibility with existing fixtures.
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:         tmpDir,
-		CacheTTL:         time.Hour,
 		NegativeTTL:      5 * time.Minute,
 		MaxBytes:         maxBytes,
 		DisableDiskCache: maxBytes == 0,

internal/imgcache/module.go

@@ -4,6 +4,5 @@ import "time"
 
 // CacheConfig defaults.
 const (
-	DefaultCacheTTL    = 24 * time.Hour
 	DefaultNegativeTTL = 5 * time.Minute
 )

internal/imgcache/negative_cache_internal_test.go

@@ -15,7 +15,6 @@ func TestNegativeCache_StoreAndCheck(t *testing.T) {
 
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    dir,
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {
@@ -63,7 +62,6 @@ func TestNegativeCache_Expired(t *testing.T) {
 
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    dir,
-		CacheTTL:    time.Hour,
 		NegativeTTL: 1 * time.Millisecond, // very short TTL
 	})
 	if err != nil {

internal/imgcache/pendingcounts.go

@@ -0,0 +1,149 @@
+package imgcache
+
+import (
+	"context"
+	"fmt"
+	"sync/atomic"
+)
+
+// StartPendingCountWrites starts the goroutine that writes the pending
+// counts to the database whenever there are some, one UPDATE at a time:
+// however many requests pass their deadline before their counts are
+// written, at most this one write waits for the database. It is a no-op
+// when already started. The goroutine outlives the caller, so it runs with
+// its own context, which StopPendingCountWrites cancels to tell it to
+// finish.
+func (c *Cache) StartPendingCountWrites() {
+	if c.pendingCountsCancel != nil {
+		return
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	c.pendingCountsCancel = cancel
+
+	go func() {
+		c.pendingCountWriteLoop(ctx)
+	}()
+}
+
+// StopPendingCountWrites tells the goroutine StartPendingCountWrites
+// started, if it did, to finish, and waits for its write under way to end
+// and for it to return. It then writes the pending counts left, so that
+// they are written before the database closes. It waits for all of this at
+// most until ctx ends, then logs the counts not written and returns an
+// error.
+func (c *Cache) StopPendingCountWrites(ctx context.Context) error {
+	var err error
+
+	if c.pendingCountsCancel != nil {
+		c.pendingCountsCancel()
+
+		select {
+		case <-c.pendingCountsDone:
+		case <-ctx.Done():
+			// The write given up on here may not reach the database:
+			// database/sql closes without waiting for a query, and a
+			// write still waiting for the connection fails when the
+			// database closes.
+			err = fmt.Errorf("pending counts still being written: %w", ctx.Err())
+		}
+	}
+
+	if err == nil {
+		err = c.writePendingCounts(ctx)
+	}
+
+	if err != nil {
+		c.log.Error("counts not written at shutdown",
+			"hits", c.pendingHits.Load(),
+			"misses", c.pendingMisses.Load(),
+			"upstream_fetches", c.pendingUpstreamFetches.Load(),
+			"upstream_fetch_bytes", c.pendingUpstreamFetchBytes.Load(),
+			"transforms", c.pendingTransforms.Load(),
+			"error", err,
+		)
+
+		return err
+	}
+
+	return nil
+}
+
+// pendingCountWriteLoop is the body of the goroutine
+// StartPendingCountWrites starts. Once ctx is cancelled it starts no write
+// and returns, but does not cut off a write under way then. A write that
+// fails leaves the counts pending, for the next write.
+func (c *Cache) pendingCountWriteLoop(ctx context.Context) {
+	defer close(c.pendingCountsDone)
+
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-c.pendingCountsAdded:
+		}
+
+		// select picks at random when ctx is cancelled and a wakeup
+		// waits too.
+		if ctx.Err() != nil {
+			return
+		}
+
+		err := c.writePendingCounts(context.WithoutCancel(ctx))
+		if err != nil {
+			c.log.Warn("failed to write pending counts", "error", err)
+		}
+	}
+}
+
+// addPendingCount adds n to pendingCount, one of the pending counts, and
+// wakes the goroutine that writes them. pendingCountsAdded has capacity one,
+// so a wakeup already waiting is enough.
+func (c *Cache) addPendingCount(pendingCount *atomic.Int64, n int64) {
+	pendingCount.Add(n)
+
+	select {
+	case c.pendingCountsAdded <- struct{}{}:
+	default:
+	}
+}
+
+// writePendingCounts adds the pending counts to the cache_stats row in one
+// UPDATE, then takes what it wrote out of them, leaving any added
+// meanwhile. When the UPDATE fails, they all stay pending.
+func (c *Cache) writePendingCounts(ctx context.Context) error {
+	c.pendingCountsWriteMutex.Lock()
+	defer c.pendingCountsWriteMutex.Unlock()
+
+	hits := c.pendingHits.Load()
+	misses := c.pendingMisses.Load()
+	upstreamFetches := c.pendingUpstreamFetches.Load()
+	upstreamFetchBytes := c.pendingUpstreamFetchBytes.Load()
+	transforms := c.pendingTransforms.Load()
+
+	if hits+misses+upstreamFetches+upstreamFetchBytes+transforms == 0 {
+		return nil
+	}
+
+	_, err := c.db.ExecContext(ctx, `
+		UPDATE cache_stats
+		SET hit_count = hit_count + ?,
+		    miss_count = miss_count + ?,
+		    upstream_fetch_count = upstream_fetch_count + ?,
+		    upstream_fetch_bytes = upstream_fetch_bytes + ?,
+		    transform_count = transform_count + ?,
+		    last_updated_at = CURRENT_TIMESTAMP
+		WHERE id = 1
+	`, hits, misses, upstreamFetches, upstreamFetchBytes, transforms)
+	if err != nil {
+		return fmt.Errorf("failed to write pending counts: %w", err)
+	}
+
+	c.pendingHits.Add(-hits)
+	c.pendingMisses.Add(-misses)
+	c.pendingUpstreamFetches.Add(-upstreamFetches)
+	c.pendingUpstreamFetchBytes.Add(-upstreamFetchBytes)
+	c.pendingTransforms.Add(-transforms)
+
+	return nil
+}

internal/imgcache/pendingcounts_internal_test.go

@@ -0,0 +1,327 @@
+package imgcache
+
+import (
+	"context"
+	"database/sql"
+	"errors"
+	"testing"
+	"time"
+)
+
+// pendingCountsWait is how long a test waits for the database's connection,
+// for a count to reach the database, for a write to start waiting for the
+// connection, for a request to return past its deadline, or for
+// StopPendingCountWrites to return.
+const pendingCountsWait = 5 * time.Second
+
+// otherWritesWait is how long a test gives other writes, once one write waits
+// for the database's connection, to start waiting for it too.
+const otherWritesWait = 100 * time.Millisecond
+
+// holdDatabase takes the one connection of cache's database, so that every
+// other query waits for it, and returns the func that frees it.
+func holdDatabase(t *testing.T, cache *Cache) func() {
+	t.Helper()
+
+	ctx, cancel := context.WithTimeout(t.Context(), pendingCountsWait)
+	defer cancel()
+
+	conn, err := cache.db.Conn(ctx)
+	if err != nil {
+		t.Fatalf("failed to take the database connection: %v", err)
+	}
+
+	return func() { _ = conn.Close() }
+}
+
+// waitForConnectionWaits waits until db.Stats().WaitCount, the number of
+// times a caller has waited for a connection, reaches waits.
+func waitForConnectionWaits(t *testing.T, db *sql.DB, waits int64) {
+	t.Helper()
+
+	deadline := time.Now().Add(pendingCountsWait)
+
+	for db.Stats().WaitCount < waits {
+		if time.Now().After(deadline) {
+			t.Fatalf("callers waited for the database connection %d times, want %d",
+				db.Stats().WaitCount, waits)
+		}
+
+		time.Sleep(10 * time.Millisecond)
+	}
+}
+
+// waitForCounters waits until the cache_stats row holds want.
+func waitForCounters(t *testing.T, cache *Cache, want cacheStatsCounters) {
+	t.Helper()
+
+	deadline := time.Now().Add(pendingCountsWait)
+
+	for {
+		got := readCacheStatsCounters(t, cache)
+		if got == want {
+			return
+		}
+
+		if time.Now().After(deadline) {
+			t.Fatalf("counters = %+v, want %+v", got, want)
+		}
+
+		time.Sleep(10 * time.Millisecond)
+	}
+}
+
+// TestService_Get_ReturnsByItsDeadlineWhileTheDatabaseIsBusy holds the
+// database's one connection while a request whose fetch is held reaches its
+// deadline. The request must return with the deadline's error while the
+// connection is still held, as it waits for no database write past its
+// deadline, and its miss must reach the database once the connection is free.
+func TestService_Get_ReturnsByItsDeadlineWhileTheDatabaseIsBusy(t *testing.T) {
+	t.Parallel()
+
+	svc, fixtures, fetcher := setupHeldFetchService(t)
+
+	svc.cache.StartPendingCountWrites()
+	defer func() { _ = svc.cache.StopPendingCountWrites(t.Context()) }()
+
+	// Room for the request to reach its fetch on a busy host
+	const timeout = 2 * time.Second
+
+	ctx, cancel := context.WithTimeout(t.Context(), timeout)
+	defer cancel()
+
+	deadline, _ := ctx.Deadline()
+
+	results := startGet(ctx, svc, photoVariant(fixtures, 85, FitCover))
+
+	// The request has made its database reads by the time it fetches.
+	select {
+	case <-fetcher.started:
+	case <-time.After(timeout):
+		t.Fatal("request did not reach its fetch by its deadline")
+	}
+
+	releaseDatabase := holdDatabase(t, svc.cache)
+	defer releaseDatabase()
+
+	select {
+	case got := <-results:
+		t.Logf("Get() returned %v after its deadline, error = %v",
+			time.Since(deadline), got.err)
+
+		if !errors.Is(got.err, context.DeadlineExceeded) {
+			t.Errorf("Get() error = %v, want %v", got.err, context.DeadlineExceeded)
+		}
+	case <-time.After(time.Until(deadline) + pendingCountsWait):
+		t.Fatal("request did not return while the database was busy")
+	}
+
+	releaseDatabase()
+
+	waitForCounters(t, svc.cache, cacheStatsCounters{missCount: 1})
+}
+
+// TestIncrementStats_ManyCountsPastTheirDeadlineLeaveOneWriteWaiting holds the
+// database's one connection while many misses are counted past their
+// deadline. At most one write may then wait for the connection, and every
+// miss must reach the database once it is free.
+func TestIncrementStats_ManyCountsPastTheirDeadlineLeaveOneWriteWaiting(t *testing.T) {
+	t.Parallel()
+
+	cache, _ := newEvictionTestCache(t, 1<<30)
+
+	cache.StartPendingCountWrites()
+	defer func() { _ = cache.StopPendingCountWrites(t.Context()) }()
+
+	releaseDatabase := holdDatabase(t, cache)
+	defer releaseDatabase()
+
+	waitsBefore := cache.db.Stats().WaitCount
+
+	// Writes given a context past its deadline fail at once.
+	ended, cancel := context.WithDeadline(t.Context(), time.Now())
+	defer cancel()
+
+	const misses = 20
+
+	for range misses {
+		cache.IncrementStats(ended, false, 0)
+	}
+
+	// Once one write waits for the connection, any others start waiting too.
+	waitForConnectionWaits(t, cache.db, waitsBefore+1)
+	time.Sleep(otherWritesWait)
+
+	waits := cache.db.Stats().WaitCount - waitsBefore
+	t.Logf("%d writes waited for the database", waits)
+
+	if waits > 1 {
+		t.Errorf("%d writes waited for the database, want at most 1", waits)
+	}
+
+	releaseDatabase()
+
+	waitForCounters(t, cache, cacheStatsCounters{missCount: misses})
+}
+
+// TestStats_IncludesPendingCounts counts hits and misses whose writes miss
+// their deadline, and checks that Stats adds them to the database's counts
+// before they are written.
+func TestStats_IncludesPendingCounts(t *testing.T) {
+	t.Parallel()
+
+	cache, _ := newEvictionTestCache(t, 1<<30)
+
+	_, err := cache.db.ExecContext(t.Context(),
+		`UPDATE cache_stats SET hit_count = 75, miss_count = 25 WHERE id = 1`)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Writes given a context past its deadline fail at once.
+	ended, cancel := context.WithDeadline(t.Context(), time.Now())
+	defer cancel()
+
+	cache.IncrementStats(ended, true, 0)
+	cache.IncrementStats(ended, true, 0)
+	cache.IncrementStats(ended, false, 0)
+
+	stats, err := cache.Stats(t.Context())
+	if err != nil {
+		t.Fatalf("Stats() error = %v", err)
+	}
+
+	if stats.HitCount != 77 || stats.MissCount != 26 {
+		t.Errorf("HitCount = %d, MissCount = %d, want 77 and 26",
+			stats.HitCount, stats.MissCount)
+	}
+
+	want := cacheStatsCounters{hitCount: 75, missCount: 25}
+	if got := readCacheStatsCounters(t, cache); got != want {
+		t.Errorf("counters in the database = %+v, want %+v", got, want)
+	}
+}
+
+// TestStopPendingCountWrites_WritesPendingCounts holds the database's one
+// connection while the goroutine that writes the pending counts waits for it,
+// counts more, then stops that goroutine. The stop must leave the write under
+// way to finish rather than cut it off, and once the connection is free it
+// must write the counts that write left.
+func TestStopPendingCountWrites_WritesPendingCounts(t *testing.T) {
+	t.Parallel()
+
+	cache, _ := newEvictionTestCache(t, 1<<30)
+
+	cache.StartPendingCountWrites()
+	defer func() { _ = cache.StopPendingCountWrites(t.Context()) }()
+
+	releaseDatabase := holdDatabase(t, cache)
+	defer releaseDatabase()
+
+	waitsBefore := cache.db.Stats().WaitCount
+
+	// Writes given a context past its deadline fail at once.
+	ended, cancel := context.WithDeadline(t.Context(), time.Now())
+	defer cancel()
+
+	cache.IncrementStats(ended, true, 0)
+
+	// The goroutine's write waits for the connection.
+	waitForConnectionWaits(t, cache.db, waitsBefore+1)
+
+	// Counted after that write read the pending counts
+	cache.IncrementStats(ended, false, 0)
+	cache.IncrementUpstreamFetch(ended, 1024)
+	cache.IncrementTransformCount(ended)
+
+	// The goroutine is told to finish here as well as by the stop, so that it
+	// is told before its write under way ends, however late the stop runs.
+	// The counts above then reach the database only through the stop's own
+	// write.
+	cache.pendingCountsCancel()
+
+	stopped := make(chan error, 1)
+
+	go func() {
+		stopped <- cache.StopPendingCountWrites(t.Context())
+	}()
+
+	// Had the stop cut off the write under way, its own write would wait for
+	// the connection too.
+	time.Sleep(otherWritesWait)
+
+	waits := cache.db.Stats().WaitCount - waitsBefore
+	t.Logf("%d writes waited for the database", waits)
+
+	if waits > 1 {
+		t.Errorf("%d writes waited for the database, want 1: "+
+			"the stop cut off the write under way", waits)
+	}
+
+	releaseDatabase()
+
+	select {
+	case err := <-stopped:
+		if err != nil {
+			t.Fatalf("StopPendingCountWrites() error = %v", err)
+		}
+	case <-time.After(pendingCountsWait):
+		t.Fatal("StopPendingCountWrites() did not return once the database was free")
+	}
+
+	want := cacheStatsCounters{1, 1, 1, 1024, 1}
+	if got := readCacheStatsCounters(t, cache); got != want {
+		t.Errorf("counters = %+v, want %+v", got, want)
+	}
+}
+
+// TestStopPendingCountWrites_ReturnsWhenItsContextEnds holds the database's
+// one connection while the goroutine that writes the pending counts waits for
+// it, then stops that goroutine with a context that ends first. The stop must
+// return the context's error, and the write under way must still reach the
+// database once the connection is free.
+func TestStopPendingCountWrites_ReturnsWhenItsContextEnds(t *testing.T) {
+	t.Parallel()
+
+	cache, _ := newEvictionTestCache(t, 1<<30)
+
+	cache.StartPendingCountWrites()
+	defer func() { _ = cache.StopPendingCountWrites(t.Context()) }()
+
+	releaseDatabase := holdDatabase(t, cache)
+	defer releaseDatabase()
+
+	waitsBefore := cache.db.Stats().WaitCount
+
+	// Writes given a context past its deadline fail at once.
+	ended, cancel := context.WithDeadline(t.Context(), time.Now())
+	defer cancel()
+
+	cache.IncrementStats(ended, true, 0)
+
+	// The goroutine's write waits for the connection.
+	waitForConnectionWaits(t, cache.db, waitsBefore+1)
+
+	stopCtx, cancelStop := context.WithCancel(t.Context())
+	cancelStop()
+
+	stopped := make(chan error, 1)
+
+	go func() {
+		stopped <- cache.StopPendingCountWrites(stopCtx)
+	}()
+
+	select {
+	case err := <-stopped:
+		if !errors.Is(err, context.Canceled) {
+			t.Errorf("StopPendingCountWrites() error = %v, want %v",
+				err, context.Canceled)
+		}
+	case <-time.After(pendingCountsWait):
+		t.Fatal("StopPendingCountWrites() did not return once its context ended")
+	}
+
+	releaseDatabase()
+
+	waitForCounters(t, cache, cacheStatsCounters{hitCount: 1})
+}

internal/imgcache/service.go

@@ -128,6 +128,12 @@ func NewService(cfg *ServiceConfig) (*Service, error) {
 // ErrNegativeCached is returned when a URL is in the negative cache (recently failed).
 var ErrNegativeCached = errors.New("request is in negative cache (recently failed)")
 
+// ErrProcessingTimedOut is returned when processing an image stopped because
+// the downstream_timeout of the request doing it ended, to that request and to
+// every request that waited for its image.
+var ErrProcessingTimedOut = errors.New(
+	"image processing stopped at the deadline of the request doing it")
+
 // Get retrieves a processed image, fetching and processing if necessary.
 func (s *Service) Get(ctx context.Context, req *ImageRequest) (*ImageResponse, error) {
 	// Propagate AllowHTTP setting to the request
@@ -162,7 +168,7 @@ func (s *Service) Get(ctx context.Context, req *ImageRequest) (*ImageResponse, e
 			// Fall through to re-process
 		} else {
 			// Counted also when the request context has ended meanwhile
-			s.cache.IncrementStats(context.WithoutCancel(ctx), true, 0)
+			s.cache.IncrementStats(ctx, true, 0)
 
 			return &ImageResponse{
 				Content:       reader,
@@ -179,7 +185,7 @@ func (s *Service) Get(ctx context.Context, req *ImageRequest) (*ImageResponse, e
 	// failed or the request context has ended meanwhile
 	response, err := s.processOrWait(ctx, req)
 
-	s.cache.IncrementStats(context.WithoutCancel(ctx), false, 0)
+	s.cache.IncrementStats(ctx, false, 0)
 
 	if err != nil {
 		return nil, err
@@ -300,14 +306,8 @@ func (s *Service) processOrWait(
 				}
 			}()
 
-			processingCtx := context.WithoutCancel(ctx)
-
-			if deadline, ok := ctx.Deadline(); ok {
-				var cancel context.CancelFunc
-
-				processingCtx, cancel = context.WithDeadline(processingCtx, deadline)
-				defer cancel()
-			}
+			processingCtx, cancel := withoutCancelKeepingDeadline(ctx)
+			defer cancel()
 
 			return s.processFromSourceOrFetch(processingCtx, req, cacheKey)
 		})
@@ -344,6 +344,20 @@ func (s *Service) processOrWait(
 	}, nil
 }
 
+// withoutCancelKeepingDeadline returns ctx without its cancellation but with
+// its deadline, if it has one, and the func that releases the returned
+// context.
+func withoutCancelKeepingDeadline(
+	ctx context.Context,
+) (context.Context, context.CancelFunc) {
+	deadline, hasDeadline := ctx.Deadline()
+	if !hasDeadline {
+		return context.WithoutCancel(ctx), func() {}
+	}
+
+	return context.WithDeadline(context.WithoutCancel(ctx), deadline)
+}
+
 // loadCachedSource opens source content from cache, without reading it, and
 // returns it with its size; nil if the cached data is unavailable, empty or
 // exceeds maxResponseSize.
@@ -453,7 +467,7 @@ func (s *Service) fetchAndProcess(
 	fetchBytes := int64(len(sourceData))
 
 	// Counted also when the request context has ended meanwhile
-	s.cache.IncrementUpstreamFetch(context.WithoutCancel(ctx), fetchBytes)
+	s.cache.IncrementUpstreamFetch(ctx, fetchBytes)
 
 	if err != nil {
 		return nil, fmt.Errorf("failed to read upstream response: %w", err)
@@ -535,7 +549,7 @@ func (s *Service) processAndStore(
 	processDuration := time.Since(processStart)
 
 	// Counted also when the request context has ended meanwhile
-	s.cache.IncrementTransformCount(context.WithoutCancel(ctx))
+	s.cache.IncrementTransformCount(ctx)
 
 	// Read processed content
 	processedData, err := io.ReadAll(processResult.Content)

internal/imgcache/service_internal_test.go

@@ -2,7 +2,9 @@ package imgcache
 
 import (
 	"context"
+	"encoding/json"
 	"io"
+	"os"
 	"testing"
 	"time"
 
@@ -430,6 +432,172 @@ func TestService_Get_Caching(t *testing.T) {
 	}
 }
 
+// TestService_Get_ServesSourceAndVariantCachedTwoDaysAgo caches a source and
+// a variant of it and sets back two days every time pixa stored for them.
+// For the source: fetched_at in source_metadata, fetched_at and
+// last_accessed_at in source_content, the fetched_at in its metadata file,
+// and the access and modification times of its file and of its metadata
+// file. For the variant: created_at and last_accessed_at in variant_content,
+// the created_at in its .meta file, and the access and modification times of
+// its file and of its .meta file. It then checks that the variant is still
+// served from the cache and that a new variant is made from the cached source
+// without fetching it again.
+//
+// Left at the present: the files' status change and birth times, which a
+// program cannot set, and the times of the directories holding the files,
+// which other entries share. The source's stored response headers are empty,
+// as the mock fetcher sends none. expires_at, etag and last_modified in
+// source_metadata and in the metadata file are never written, nor are the
+// request_cache and output_content tables.
+func TestService_Get_ServesSourceAndVariantCachedTwoDaysAgo(t *testing.T) {
+	t.Parallel()
+
+	svc, fixtures := SetupTestService(t)
+	variant := photoVariant(fixtures, 85, FitCover)
+
+	resp, err := svc.Get(t.Context(), variant)
+	if err != nil {
+		t.Fatalf("Get() first request error = %v", err)
+	}
+
+	_ = resp.Content.Close()
+
+	twoDaysAgo := time.Now().AddDate(0, 0, -2)
+	backdateSource(t, svc.cache, variant, twoDaysAgo)
+	backdateVariant(t, svc.cache, variant, twoDaysAgo)
+
+	resp, err = svc.Get(t.Context(), variant)
+	if err != nil {
+		t.Fatalf("Get() of the cached variant error = %v", err)
+	}
+
+	_ = resp.Content.Close()
+
+	if resp.CacheStatus != CacheHit {
+		t.Errorf("cached variant CacheStatus = %q, want %q",
+			resp.CacheStatus, CacheHit)
+	}
+
+	// Another quality is a new variant, made from the cached source
+	resp, err = svc.Get(t.Context(), photoVariant(fixtures, 40, FitCover))
+	if err != nil {
+		t.Fatalf("Get() of a new variant error = %v", err)
+	}
+
+	_ = resp.Content.Close()
+
+	fetches := readCacheStatsCounters(t, svc.cache).upstreamFetchCount
+	if fetches != 1 {
+		t.Errorf("upstream fetches = %d, want 1", fetches)
+	}
+}
+
+// backdateSource sets every time pixa stored for req's cached source to when,
+// as listed on TestService_Get_ServesSourceAndVariantCachedTwoDaysAgo.
+func backdateSource(
+	t *testing.T, cache *Cache, req *ImageRequest, when time.Time,
+) {
+	t.Helper()
+
+	pathHash := HashPath(req.SourcePath + "?" + req.SourceQuery)
+
+	meta, err := cache.srcMetadata.Load(req.SourceHost, pathHash)
+	if err != nil {
+		t.Fatalf("reading the source's metadata file: %v", err)
+	}
+
+	meta.FetchedAt = when.Unix()
+
+	err = cache.srcMetadata.Store(req.SourceHost, pathHash, meta)
+	if err != nil {
+		t.Fatalf("writing the source's metadata file: %v", err)
+	}
+
+	dbTime := when.UTC().Format(sqliteTimestampFormat)
+
+	updateOneRow(t, cache, `
+		UPDATE source_metadata SET fetched_at = ?
+		WHERE source_host = ? AND source_path = ? AND source_query = ?
+	`, dbTime, req.SourceHost, req.SourcePath, req.SourceQuery)
+	updateOneRow(t, cache,
+		`UPDATE source_content SET fetched_at = ? WHERE content_hash = ?`,
+		dbTime, meta.ContentHash)
+	setSourceLastAccessed(t, cache, ContentHash(meta.ContentHash), when)
+
+	for _, path := range []string{
+		cache.srcContent.hashToPath(ContentHash(meta.ContentHash)),
+		cache.srcMetadata.metaPath(req.SourceHost, pathHash),
+	} {
+		err = os.Chtimes(path, when, when)
+		if err != nil {
+			t.Fatalf("setting back the times of %s: %v", path, err)
+		}
+	}
+}
+
+// backdateVariant sets every time pixa stored for the cached variant req asks
+// for to when, as listed on
+// TestService_Get_ServesSourceAndVariantCachedTwoDaysAgo.
+func backdateVariant(
+	t *testing.T, cache *Cache, req *ImageRequest, when time.Time,
+) {
+	t.Helper()
+
+	key := CacheKey(req)
+
+	content, size, contentType, err := cache.variants.LoadWithMeta(key)
+	if err != nil || contentType == "" {
+		t.Fatalf("reading the variant's .meta file: content type %q, error %v",
+			contentType, err)
+	}
+
+	_ = content.Close()
+
+	meta, err := json.Marshal(VariantMeta{
+		ContentType: contentType,
+		Size:        size,
+		CreatedAt:   when.Unix(),
+	})
+	if err != nil {
+		t.Fatalf("encoding the variant's .meta file: %v", err)
+	}
+
+	path := cache.variants.keyToPath(key)
+
+	err = os.WriteFile(path+variantMetaSuffix, meta, StorageFilePerm)
+	if err != nil {
+		t.Fatalf("writing the variant's .meta file: %v", err)
+	}
+
+	updateOneRow(t, cache,
+		`UPDATE variant_content SET created_at = ? WHERE cache_key = ?`,
+		when.UTC().Format(sqliteTimestampFormat), string(key))
+	setVariantLastAccessed(t, cache, key, when)
+
+	for _, file := range []string{path, path + variantMetaSuffix} {
+		err = os.Chtimes(file, when, when)
+		if err != nil {
+			t.Fatalf("setting back the times of %s: %v", file, err)
+		}
+	}
+}
+
+// updateOneRow runs the UPDATE statement query and fails the test unless it
+// changed exactly one row.
+func updateOneRow(t *testing.T, cache *Cache, query string, args ...any) {
+	t.Helper()
+
+	result, err := cache.db.ExecContext(t.Context(), query, args...)
+	if err != nil {
+		t.Fatalf("%s: %v", query, err)
+	}
+
+	rows, err := result.RowsAffected()
+	if err != nil || rows != 1 {
+		t.Fatalf("%s: rows = %d, error = %v, want 1 row", query, rows, err)
+	}
+}
+
 func TestService_Get_DifferentSizes(t *testing.T) {
 	t.Parallel()
 

internal/imgcache/stats_internal_test.go

@@ -43,7 +43,6 @@ func TestStats_HitRateIsRatio(t *testing.T) {
 
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    dir,
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {
@@ -91,7 +90,6 @@ func TestStats_ZeroCounts(t *testing.T) {
 
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    dir,
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {
@@ -120,7 +118,6 @@ func TestStats_LogsFailedCountQueries(t *testing.T) {
 
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    t.TempDir(),
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 		Logger:      slog.New(slog.NewJSONHandler(&logBuf, nil)),
 	})
@@ -162,7 +159,6 @@ func TestIncrementStats_LogsFailedUpdates(t *testing.T) {
 
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    t.TempDir(),
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 		Logger:      slog.New(slog.NewJSONHandler(&logBuf, nil)),
 	})
@@ -232,7 +228,6 @@ func TestStats_DisabledCacheReportsNoItems(t *testing.T) {
 
 	disabled, err := NewCache(enabled.db, CacheConfig{
 		StateDir:         t.TempDir(),
-		CacheTTL:         time.Hour,
 		NegativeTTL:      5 * time.Minute,
 		DisableDiskCache: true,
 	})

internal/imgcache/storage.go

@@ -10,6 +10,7 @@ import (
 	"log/slog"
 	"os"
 	"path/filepath"
+	"strings"
 	"time"
 )
 
@@ -371,11 +372,13 @@ func HashPath(path string) PathHash {
 
 // CacheKey generates a unique key for a request variant.
 // Format: sha256(host:path:query:width:height:format:quality:fit_mode)
+// The path and query go through escapePercentAndColon, so that a ":" in
+// either cannot be taken for the one between them.
 func CacheKey(req *ImageRequest) VariantKey {
 	data := fmt.Sprintf("%s:%s:%s:%d:%d:%s:%d:%s",
 		req.SourceHost,
-		req.SourcePath,
-		req.SourceQuery,
+		escapePercentAndColon(req.SourcePath),
+		escapePercentAndColon(req.SourceQuery),
 		req.Size.Width,
 		req.Size.Height,
 		req.Format,
@@ -387,6 +390,14 @@ func CacheKey(req *ImageRequest) VariantKey {
 	return VariantKey(hex.EncodeToString(h[:]))
 }
 
+// escapePercentAndColon returns s with each "%" written as "%25" and each
+// ":" as "%3A", so that the result holds no ":".
+func escapePercentAndColon(s string) string {
+	s = strings.ReplaceAll(s, "%", "%25")
+
+	return strings.ReplaceAll(s, ":", "%3A")
+}
+
 // VariantStorage handles key-based file storage for processed image variants.
 // Files are stored at: <basedir>/<ab>/<cd>/<cache_key>
 // Metadata is stored at: <basedir>/<ab>/<cd>/<cache_key>.meta

internal/imgcache/storage_internal_test.go

@@ -407,6 +407,62 @@ func TestCacheKey(t *testing.T) {
 	}
 }
 
+// TestCacheKey_ColonOrPercentInPathOrQuery checks that two different source
+// URLs whose path or query holds ":" or "%" get different keys.
+func TestCacheKey_ColonOrPercentInPathOrQuery(t *testing.T) {
+	t.Parallel()
+
+	cases := []struct {
+		name          string
+		path1, query1 string
+		path2, query2 string
+	}{
+		{
+			name:  "colon moved from path to query",
+			path1: "/a:b", query1: "",
+			path2: "/a", query2: "b:",
+		},
+		{
+			name:  "colon in path and its percent-encoding",
+			path1: "/a:b", query1: "",
+			path2: "/a%3Ab", query2: "",
+		},
+	}
+
+	for _, tt := range cases {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			key1 := CacheKey(&ImageRequest{
+				SourceHost:  testHostCDN,
+				SourcePath:  tt.path1,
+				SourceQuery: tt.query1,
+				Size:        Size{Width: 800, Height: 600},
+				Format:      FormatWebP,
+				Quality:     85,
+				FitMode:     FitCover,
+			})
+			key2 := CacheKey(&ImageRequest{
+				SourceHost:  testHostCDN,
+				SourcePath:  tt.path2,
+				SourceQuery: tt.query2,
+				Size:        Size{Width: 800, Height: 600},
+				Format:      FormatWebP,
+				Quality:     85,
+				FitMode:     FitCover,
+			})
+
+			t.Logf("CacheKey() of path %q query %q = %s", tt.path1, tt.query1, key1)
+			t.Logf("CacheKey() of path %q query %q = %s", tt.path2, tt.query2, key2)
+
+			if key1 == key2 {
+				t.Errorf("CacheKey() gave path %q query %q and path %q query %q "+
+					"the same key", tt.path1, tt.query1, tt.path2, tt.query2)
+			}
+		})
+	}
+}
+
 // TestVariantStorage_StoreLogsFailedMetaWrite verifies that a .meta write
 // that fails is logged at warn and the store still succeeds.
 func TestVariantStorage_StoreLogsFailedMetaWrite(t *testing.T) {

internal/imgcache/testutil_internal_test.go

@@ -185,7 +185,6 @@ func SetupTestService(
 
 	cache, err := NewCache(db, CacheConfig{
 		StateDir:    tmpDir,
-		CacheTTL:    time.Hour,
 		NegativeTTL: 5 * time.Minute,
 	})
 	if err != nil {

internal/imgcache/urlparser.go

@@ -7,6 +7,8 @@ import (
 	"regexp"
 	"strconv"
 	"strings"
+
+	"sneak.berlin/go/pixa/internal/imageprocessor"
 )
 
 // URL parsing errors.
@@ -20,8 +22,9 @@ var (
 	ErrPathTraversal     = errors.New("path traversal detected")
 )
 
-// MaxDimension is the maximum allowed width or height.
-const MaxDimension = 8192
+// MaxDimension is the maximum allowed width or height, the same as the image
+// processor's maximum for an output image.
+const MaxDimension = imageprocessor.MaxOutputDimension
 
 // ValidateDimension checks one requested width or height; name ("width" or
 // "height") appears in the error. 0 means "original size" and is valid.

internal/middleware/middleware.go

@@ -98,9 +98,10 @@ func (s *Middleware) ClientIP() func(http.Handler) http.Handler {
 // holds whole; an IPv4-mapped address (::ffff:a.b.c.d) is counted as the
 // IPv4 address it carries, since every such address falls in the same /64.
 // Counts are kept only for the current and the previous window, so memory
-// stays bounded.
+// stays bounded. limitHandler writes the 429 after Retry-After is set; nil
+// keeps the rate limit library's plain-text "Too Many Requests".
 func (s *Middleware) RateLimit(
-	requestLimit int, window time.Duration,
+	requestLimit int, window time.Duration, limitHandler http.HandlerFunc,
 ) func(http.Handler) http.Handler {
 	return httprate.LimitBy(requestLimit, window,
 		func(r *http.Request) (string, error) {
@@ -112,7 +113,8 @@ func (s *Middleware) RateLimit(
 			}
 
 			return httprate.CanonicalizeIP(ip), nil
-		})
+		},
+		httprate.WithLimitHandler(limitHandler))
 }
 
 // requestIDPattern is what a request's own X-Request-Id must look like to be

internal/server/computed_side_internal_test.go

@@ -0,0 +1,153 @@
+package server
+
+import (
+	"bytes"
+	"encoding/json"
+	"image/jpeg"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"sneak.berlin/go/pixa/internal/encurl"
+	"sneak.berlin/go/pixa/internal/imgcache"
+)
+
+// TestImageProxyRefusesComputedSideAboveMaximum requests sizes of a 3000x600
+// and a 600x3000 PNG through both image routes. A size with one side 0, whose
+// side computed from the source's proportions would be above 8192, is refused
+// with 400 and the JSON error body. The sizes after those refusals are served
+// as before, which also shows that a refusal is not cached as a failed fetch of
+// the source: orig, a size with both sides given, and a size with one side 0
+// whose computed side stays within 8192.
+func TestImageProxyRefusesComputedSideAboveMaximum(t *testing.T) {
+	t.Parallel()
+
+	// The upstream paths of the two sources
+	const (
+		widePath = "/wide.png"
+		tallPath = "/tall.png"
+	)
+
+	wide := encodeTestPNG(t, 3000, 600)
+	tall := encodeTestPNG(t, 600, 3000)
+
+	upstream := httptest.NewServer(http.HandlerFunc(
+		func(w http.ResponseWriter, r *http.Request) {
+			w.Header().Set("Content-Type", "image/png")
+
+			if r.URL.Path == tallPath {
+				_, _ = w.Write(tall)
+
+				return
+			}
+
+			_, _ = w.Write(wide)
+		}))
+	t.Cleanup(upstream.Close)
+
+	s, _, _ := startImageProxy(t, upstream)
+
+	generator, err := encurl.NewGenerator(testSigningKey)
+	if err != nil {
+		t.Fatalf("encurl.NewGenerator: %v", err)
+	}
+
+	tests := []struct {
+		path string
+		// size is the size of the /v1/image/ URL, and width and height are
+		// those of the encrypted URL
+		size          string
+		width, height int
+		// wantWidth and wantHeight are the served image's, 0 for a refusal
+		wantWidth, wantHeight int
+	}{
+		{widePath, "0x8192", 0, 8192, 0, 0},
+		{tallPath, "8192x0", 8192, 0, 0, 0},
+		{widePath, "orig", 0, 0, 3000, 600},
+		{widePath, "300x200", 300, 200, 300, 200},
+		{widePath, "0x300", 0, 300, 1500, 300},
+		{tallPath, "300x0", 300, 0, 300, 1500},
+	}
+
+	for _, tt := range tests {
+		token, err := generator.Generate(&encurl.Payload{
+			SourceHost: upstreamHost, SourcePath: tt.path,
+			Width: tt.width, Height: tt.height, Format: imgcache.FormatJPEG,
+		})
+		if err != nil {
+			t.Fatalf("encurl Generate: %v", err)
+		}
+
+		for _, target := range []string{
+			"/v1/image/" + upstreamHost + tt.path + "/" + tt.size + ".jpeg",
+			"/v1/e/" + token + "/img.jpg",
+		} {
+			rec := httptest.NewRecorder()
+			s.ServeHTTP(rec, httptest.NewRequestWithContext(
+				t.Context(), http.MethodGet, target, nil))
+			t.Logf("GET %s: %d, Content-Type %s",
+				target, rec.Code, rec.Header().Get("Content-Type"))
+
+			if tt.wantWidth == 0 {
+				checkSizeRefused(t, rec)
+			} else {
+				checkServedSize(t, rec, tt.wantWidth, tt.wantHeight)
+			}
+		}
+	}
+}
+
+// checkSizeRefused checks that rec answers 400 with the JSON error body that
+// refuses a size whose side computed from the source's proportions is above
+// 8192.
+func checkSizeRefused(t *testing.T, rec *httptest.ResponseRecorder) {
+	t.Helper()
+
+	const wantError = "size too large for the source: " +
+		"the side computed from its proportions is above 8192"
+
+	var body struct {
+		Error  string `json:"error"`
+		Status int    `json:"status"`
+	}
+
+	err := json.Unmarshal(rec.Body.Bytes(), &body)
+	if err != nil {
+		t.Errorf("status %d, body is not JSON: %v", rec.Code, err)
+
+		return
+	}
+
+	if rec.Code != http.StatusBadRequest || body.Status != http.StatusBadRequest ||
+		body.Error != wantError {
+		t.Errorf("status %d, body %+v, want %d and the error %q",
+			rec.Code, body, http.StatusBadRequest, wantError)
+	}
+}
+
+// checkServedSize checks that rec answers 200 with a JPEG of wantWidth by
+// wantHeight.
+func checkServedSize(
+	t *testing.T, rec *httptest.ResponseRecorder, wantWidth, wantHeight int,
+) {
+	t.Helper()
+
+	if rec.Code != http.StatusOK {
+		t.Errorf("status %d, want %d; body %s",
+			rec.Code, http.StatusOK, rec.Body.String())
+
+		return
+	}
+
+	served, err := jpeg.DecodeConfig(bytes.NewReader(rec.Body.Bytes()))
+	if err != nil {
+		t.Errorf("decoding the served JPEG: %v", err)
+
+		return
+	}
+
+	if served.Width != wantWidth || served.Height != wantHeight {
+		t.Errorf("served %dx%d, want %dx%d",
+			served.Width, served.Height, wantWidth, wantHeight)
+	}
+}

internal/server/fit_outside_internal_test.go

@@ -0,0 +1,91 @@
+package server
+
+import (
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"sneak.berlin/go/pixa/internal/encurl"
+	"sneak.berlin/go/pixa/internal/imgcache"
+)
+
+// TestImageProxyRefusesFitOutsideAboveMaximum requests sizes with fit=outside
+// through both image routes. fit=outside scales the source to cover the size,
+// and so past it on one side: a size for which that side would be above 8192
+// is refused with 400 and the JSON error body, whether it is the width or the
+// height, and a size for which it stays within 8192 is served at the scaled
+// size.
+func TestImageProxyRefusesFitOutsideAboveMaximum(t *testing.T) {
+	t.Parallel()
+
+	// The upstream paths of the sources
+	const (
+		widePath   = "/wide.png"
+		rowPath    = "/row.png"
+		columnPath = "/column.png"
+	)
+
+	sources := map[string][]byte{
+		widePath:   encodeTestPNG(t, 3000, 600),
+		rowPath:    encodeTestPNG(t, 8192, 1),
+		columnPath: encodeTestPNG(t, 1, 8192),
+	}
+
+	upstream := httptest.NewServer(http.HandlerFunc(
+		func(w http.ResponseWriter, r *http.Request) {
+			w.Header().Set("Content-Type", "image/png")
+			_, _ = w.Write(sources[r.URL.Path])
+		}))
+	t.Cleanup(upstream.Close)
+
+	s, _, _ := startImageProxy(t, upstream)
+
+	generator, err := encurl.NewGenerator(testSigningKey)
+	if err != nil {
+		t.Fatalf("encurl.NewGenerator: %v", err)
+	}
+
+	tests := []struct {
+		path          string
+		width, height int
+		// wantWidth and wantHeight are the served image's, 0 for a refusal
+		wantWidth, wantHeight int
+	}{
+		{widePath, 1, 8192, 0, 0},
+		{rowPath, 1, 8192, 0, 0},
+		{columnPath, 8192, 1, 0, 0},
+		{widePath, 1, 1200, 6000, 1200},
+	}
+
+	for _, tt := range tests {
+		token, err := generator.Generate(&encurl.Payload{
+			SourceHost: upstreamHost, SourcePath: tt.path,
+			Width: tt.width, Height: tt.height, Format: imgcache.FormatJPEG,
+			FitMode: imgcache.FitOutside,
+		})
+		if err != nil {
+			t.Fatalf("encurl Generate: %v", err)
+		}
+
+		size := fmt.Sprintf("%dx%d", tt.width, tt.height)
+
+		for _, target := range []string{
+			"/v1/image/" + upstreamHost + tt.path + "/" + size +
+				".jpeg?fit=outside",
+			"/v1/e/" + token + "/img.jpg",
+		} {
+			rec := httptest.NewRecorder()
+			s.ServeHTTP(rec, httptest.NewRequestWithContext(
+				t.Context(), http.MethodGet, target, nil))
+			t.Logf("GET %s: %d, Content-Type %s",
+				target, rec.Code, rec.Header().Get("Content-Type"))
+
+			if tt.wantWidth == 0 {
+				checkSizeRefused(t, rec)
+			} else {
+				checkServedSize(t, rec, tt.wantWidth, tt.wantHeight)
+			}
+		}
+	}
+}

internal/server/image_rate_limit_internal_test.go

@@ -0,0 +1,204 @@
+package server
+
+import (
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"strconv"
+	"testing"
+
+	"sneak.berlin/go/pixa/internal/config"
+)
+
+// requestImageFrom sends method path to s from remoteAddr and returns the
+// response. The image handlers refuse an unsigned image URL with 401 and a
+// token they cannot decrypt with 400, so either status shows the limit let
+// the request through.
+func requestImageFrom(
+	t *testing.T, s *Server, method, path, remoteAddr string,
+) *httptest.ResponseRecorder {
+	t.Helper()
+
+	req := httptest.NewRequestWithContext(t.Context(), method, path, nil)
+	req.RemoteAddr = remoteAddr
+
+	rec := httptest.NewRecorder()
+	s.ServeHTTP(rec, req)
+
+	return rec
+}
+
+// TestImageRequestLimitRefusesRequestOverLimit verifies that GET and HEAD on
+// both image routes count toward one limit per client, and that the request
+// after image_requests_per_minute of them is refused with 429 and a
+// Retry-After header.
+func TestImageRequestLimitRefusesRequestOverLimit(t *testing.T) {
+	t.Parallel()
+
+	requests := []struct {
+		method string
+		path   string
+		want   int
+	}{
+		{http.MethodGet, unsignedImagePath, http.StatusUnauthorized},
+		{http.MethodHead, unsignedImagePath, http.StatusUnauthorized},
+		{http.MethodGet, encryptedImagePath, http.StatusBadRequest},
+	}
+
+	s := newTestServer(t)
+	s.config.ImageRequestsPerMinute = len(requests)
+	s.SetupRoutes()
+
+	for _, tc := range requests {
+		rec := requestImageFrom(t, s, tc.method, tc.path, firstClient)
+		if rec.Code != tc.want {
+			t.Fatalf("%s %s within the limit status = %d, want %d",
+				tc.method, tc.path, rec.Code, tc.want)
+		}
+	}
+
+	rec := requestImageFrom(t, s, http.MethodHead, encryptedImagePath, firstClient)
+	if rec.Code != http.StatusTooManyRequests {
+		t.Fatalf("request over the limit status = %d, want %d",
+			rec.Code, http.StatusTooManyRequests)
+	}
+
+	retryAfter := rec.Header().Get("Retry-After")
+
+	seconds, err := strconv.Atoi(retryAfter)
+	if err != nil || seconds <= 0 {
+		t.Errorf("Retry-After = %q, want a positive number of seconds",
+			retryAfter)
+	}
+}
+
+// TestImageRequestLimitAnswersWithJSONError verifies that the 429 for a
+// request over the limit carries Retry-After and the JSON error body the
+// image routes send for their other errors.
+func TestImageRequestLimitAnswersWithJSONError(t *testing.T) {
+	t.Parallel()
+
+	s := newTestServer(t)
+	s.config.ImageRequestsPerMinute = 1
+	s.SetupRoutes()
+
+	requestImageFrom(t, s, http.MethodGet, unsignedImagePath, firstClient)
+
+	rec := requestImageFrom(t, s, http.MethodGet, unsignedImagePath, firstClient)
+	t.Logf("status %d, body %s", rec.Code, rec.Body.String())
+
+	if rec.Code != http.StatusTooManyRequests {
+		t.Fatalf("request over the limit status = %d, want %d",
+			rec.Code, http.StatusTooManyRequests)
+	}
+
+	retryAfter := rec.Header().Get("Retry-After")
+
+	seconds, err := strconv.Atoi(retryAfter)
+	if err != nil || seconds <= 0 {
+		t.Errorf("Retry-After = %q, want a positive number of seconds",
+			retryAfter)
+	}
+
+	contentType := rec.Header().Get("Content-Type")
+	if contentType != "application/json" {
+		t.Errorf("Content-Type = %q, want %q", contentType, "application/json")
+	}
+
+	var body struct {
+		Error     string `json:"error"`
+		Status    int    `json:"status"`
+		Timestamp string `json:"timestamp"`
+	}
+
+	err = json.NewDecoder(rec.Body).Decode(&body)
+	if err != nil {
+		t.Fatalf("body is not JSON: %v", err)
+	}
+
+	if body.Error == "" || body.Status != http.StatusTooManyRequests ||
+		body.Timestamp == "" {
+		t.Errorf("body = %+v, want an error, status %d and a timestamp",
+			body, http.StatusTooManyRequests)
+	}
+}
+
+// TestImageRequestLimitCountsClientsSeparately verifies that one client going
+// over the limit does not limit another.
+func TestImageRequestLimitCountsClientsSeparately(t *testing.T) {
+	t.Parallel()
+
+	s := newTestServer(t)
+	s.config.ImageRequestsPerMinute = 1
+	s.SetupRoutes()
+
+	rec := requestImageFrom(t, s, http.MethodGet, unsignedImagePath, firstClient)
+	if rec.Code != http.StatusUnauthorized {
+		t.Fatalf("first client's first request status = %d, want %d",
+			rec.Code, http.StatusUnauthorized)
+	}
+
+	rec = requestImageFrom(t, s, http.MethodGet, unsignedImagePath, firstClient)
+	if rec.Code != http.StatusTooManyRequests {
+		t.Fatalf("first client's second request status = %d, want %d",
+			rec.Code, http.StatusTooManyRequests)
+	}
+
+	rec = requestImageFrom(t, s, http.MethodGet, unsignedImagePath, secondClient)
+	if rec.Code != http.StatusUnauthorized {
+		t.Errorf("second client's first request status = %d, want %d",
+			rec.Code, http.StatusUnauthorized)
+	}
+}
+
+// TestImageRequestLimitZeroLetsEveryRequestThrough verifies that with
+// image_requests_per_minute set to 0 no image request is refused with 429,
+// not even past the default limit.
+func TestImageRequestLimitZeroLetsEveryRequestThrough(t *testing.T) {
+	t.Parallel()
+
+	s := newTestServer(t)
+	s.config.ImageRequestsPerMinute = 0
+	s.SetupRoutes()
+
+	for request := range config.DefaultImageRequestsPerMinute + 1 {
+		rec := requestImageFrom(t, s, http.MethodGet, unsignedImagePath, firstClient)
+		if rec.Code != http.StatusUnauthorized {
+			t.Fatalf("request %d status = %d, want %d",
+				request+1, rec.Code, http.StatusUnauthorized)
+		}
+	}
+}
+
+// TestImageRequestLimitDoesNotCountPreflight verifies that a browser's CORS
+// preflight request to an image route is answered without being counted, so
+// a client that has sent several still gets its image request through.
+func TestImageRequestLimitDoesNotCountPreflight(t *testing.T) {
+	t.Parallel()
+
+	s := newTestServer(t)
+	s.config.AccessControlAllowOrigin = config.DefaultAccessControlAllowOrigin
+	s.config.ImageRequestsPerMinute = 1
+	s.SetupRoutes()
+
+	for range 3 {
+		req := httptest.NewRequestWithContext(
+			t.Context(), http.MethodOptions, unsignedImagePath, nil)
+		req.RemoteAddr = firstClient
+		req.Header.Set("Origin", "https://app.example.com")
+		req.Header.Set("Access-Control-Request-Method", http.MethodGet)
+
+		rec := httptest.NewRecorder()
+		s.ServeHTTP(rec, req)
+
+		if rec.Code != http.StatusOK {
+			t.Fatalf("preflight status = %d, want %d", rec.Code, http.StatusOK)
+		}
+	}
+
+	rec := requestImageFrom(t, s, http.MethodGet, unsignedImagePath, firstClient)
+	if rec.Code != http.StatusUnauthorized {
+		t.Errorf("image request after preflights status = %d, want %d",
+			rec.Code, http.StatusUnauthorized)
+	}
+}

internal/server/routes.go

@@ -66,7 +66,7 @@ func (s *Server) SetupRoutes() {
 		r.Use(s.h.LimitBody(handlers.MaxFormBytes))
 		r.Use(s.h.CSRF())
 		r.Get("/", s.h.HandleRoot())
-		r.With(s.mw.RateLimit(LoginAttemptsPerMinute, time.Minute)).
+		r.With(s.mw.RateLimit(LoginAttemptsPerMinute, time.Minute, nil)).
 			Post("/", s.h.HandleRoot())
 		r.Post("/generate", s.h.HandleGenerateURL())
 	})
@@ -88,6 +88,15 @@ func (s *Server) SetupRoutes() {
 		r.Group(func(r chi.Router) {
 			r.Use(s.refuseDuringMaintenance)
 
+			// Limited per client, as the login POST is, but the 429 has
+			// the image routes' JSON error body; 0 turns the limit off.
+			// The CORS middleware answers a preflight before it gets
+			// here, so a preflight is not counted.
+			if s.config.ImageRequestsPerMinute > 0 {
+				r.Use(s.mw.RateLimit(s.config.ImageRequestsPerMinute,
+					time.Minute, s.h.HandleTooManyRequests()))
+			}
+
 			// Main image proxy route
 			// /v1/image/<host>/<path>/<width>x<height>.<format>, or with no
 			// format /v1/image/<host>/<path>/<width>x<height>

internal/server/svg_as_avif_internal_test.go

@@ -0,0 +1,71 @@
+package server
+
+import (
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	"sneak.berlin/go/pixa/internal/encurl"
+	"sneak.berlin/go/pixa/internal/imgcache"
+)
+
+// svgServedAsAVIF is an SVG whose first bytes pass the first-bytes check for
+// AVIF: its XML comment puts "ftypavis" at bytes 4 to 11, where an AVIF file
+// has its ftyp box and brand. libvips would load it as SVG.
+const svgServedAsAVIF = `<!--ftypavis--><svg xmlns="http://www.w3.org/2000/svg" ` +
+	`width="64" height="48"><rect width="64" height="48"/></svg>`
+
+// TestImageProxyRefusesSVGServedAsAVIF requests svgServedAsAVIF, which the
+// upstream answers as image/avif, through both image routes, resized as PNG
+// and in its own format (orig). Each request is answered 415, as for an SVG
+// source, never with an image.
+func TestImageProxyRefusesSVGServedAsAVIF(t *testing.T) {
+	t.Parallel()
+
+	upstream := httptest.NewServer(http.HandlerFunc(
+		func(w http.ResponseWriter, _ *http.Request) {
+			w.Header().Set("Content-Type", "image/avif")
+			_, _ = io.WriteString(w, svgServedAsAVIF)
+		}))
+	t.Cleanup(upstream.Close)
+
+	s, _, _ := startImageProxy(t, upstream)
+
+	generator, err := encurl.NewGenerator(testSigningKey)
+	if err != nil {
+		t.Fatalf("encurl.NewGenerator: %v", err)
+	}
+
+	encryptedURL := func(format imgcache.ImageFormat) string {
+		token, err := generator.Generate(&encurl.Payload{
+			SourceHost: upstreamHost, SourcePath: "/drawing.avif",
+			Width: 32, Height: 24, Format: format,
+		})
+		if err != nil {
+			t.Fatalf("encurl Generate: %v", err)
+		}
+
+		return "/v1/e/" + token + "/drawing." + string(format)
+	}
+
+	for _, target := range []string{
+		"/v1/image/" + upstreamHost + "/drawing.avif/32x24.png",
+		"/v1/image/" + upstreamHost + "/drawing.avif/64x48.orig",
+		encryptedURL(imgcache.FormatPNG),
+		encryptedURL(imgcache.FormatOriginal),
+	} {
+		rec := httptest.NewRecorder()
+		s.ServeHTTP(rec, httptest.NewRequestWithContext(
+			t.Context(), http.MethodGet, target, nil))
+		t.Logf("GET %s: %d, Content-Type %s, body %q",
+			target, rec.Code, rec.Header().Get("Content-Type"), rec.Body.String())
+
+		if rec.Code != http.StatusUnsupportedMediaType ||
+			!strings.Contains(rec.Body.String(), "unsupported source format") {
+			t.Errorf("GET %s: %d %q, want %d unsupported source format",
+				target, rec.Code, rec.Body.String(), http.StatusUnsupportedMediaType)
+		}
+	}
+}

internal/server/svg_internal_test.go

@@ -0,0 +1,41 @@
+package server
+
+import (
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+)
+
+// TestImageProxyRefusesSVG requests an image the upstream answers as SVG,
+// in its own format (orig) and resized as PNG, through pixa's router,
+// upstream fetcher and image processor. pixa does not accept SVG as a
+// source, so each request is answered 415, never with an image.
+func TestImageProxyRefusesSVG(t *testing.T) {
+	t.Parallel()
+
+	upstream := httptest.NewServer(http.HandlerFunc(
+		func(w http.ResponseWriter, _ *http.Request) {
+			w.Header().Set("Content-Type", "image/svg+xml")
+			_, _ = io.WriteString(w, `<svg xmlns="http://www.w3.org/2000/svg" `+
+				`width="64" height="48"><rect width="64" height="48"/></svg>`)
+		}))
+	t.Cleanup(upstream.Close)
+
+	s, _, _ := startImageProxy(t, upstream)
+
+	for _, sizeFormat := range []string{"64x48.orig", "32x24.png"} {
+		target := "/v1/image/" + upstreamHost + "/drawing.svg/" + sizeFormat
+
+		rec := httptest.NewRecorder()
+		s.ServeHTTP(rec, httptest.NewRequestWithContext(
+			t.Context(), http.MethodGet, target, nil))
+		t.Logf("GET %s: %d, Content-Type %s, body %s",
+			target, rec.Code, rec.Header().Get("Content-Type"), rec.Body.String())
+
+		if rec.Code != http.StatusUnsupportedMediaType {
+			t.Errorf("GET %s status = %d, want %d",
+				target, rec.Code, http.StatusUnsupportedMediaType)
+		}
+	}
+}

internal/signature/escaping_test.go

@@ -0,0 +1,113 @@
+package signature_test
+
+import (
+	"errors"
+	"testing"
+	"time"
+
+	"sneak.berlin/go/pixa/internal/signature"
+)
+
+// TestSigner_Verify_ColonCannotMovePathQueryBoundary checks that a URL
+// signed for path "/a:b" with no query does not verify for path "/a" with
+// query "b:", a different resource on the same host.
+func TestSigner_Verify_ColonCannotMovePathQueryBoundary(t *testing.T) {
+	t.Parallel()
+
+	signer := signature.New(goldenSigningKey)
+
+	signed := &signature.Request{
+		SourceHost: testHost,
+		SourcePath: "/a:b",
+		Width:      800,
+		Height:     600,
+		Format:     testFormatWebP,
+		Quality:    85,
+		FitMode:    testFitCover,
+		Expires:    time.Now().Add(1 * time.Hour),
+	}
+	signed.Signature = signer.Sign(signed)
+
+	replayed := *signed
+	replayed.SourcePath = "/a"
+	replayed.SourceQuery = "b:"
+
+	t.Logf("signature for path %q query %q: %s",
+		signed.SourcePath, signed.SourceQuery, signed.Signature)
+	t.Logf("signature for path %q query %q: %s",
+		replayed.SourcePath, replayed.SourceQuery, signer.Sign(&replayed))
+
+	err := signer.Verify(&replayed)
+	if !errors.Is(err, signature.ErrInvalid) {
+		t.Errorf("Verify() of path %q query %q with the signature for "+
+			"path %q = %v, want %v",
+			replayed.SourcePath, replayed.SourceQuery, signed.SourcePath,
+			err, signature.ErrInvalid)
+	}
+}
+
+// TestSigner_EscapedGoldenVectors pins the signature of requests whose path
+// or query holds ":" or "%", which are signed with "%" written as "%25" and
+// ":" as "%3A". The expected values were computed once, outside pixa, from
+// the signed data in each comment.
+func TestSigner_EscapedGoldenVectors(t *testing.T) {
+	t.Parallel()
+
+	signer := signature.New(goldenSigningKey)
+
+	vectors := []struct {
+		name          string
+		path          string
+		query         string
+		wantSignature string
+	}{
+		{
+			name: "colon in path",
+			path: "/photos/a:b.jpg",
+			// Signed data:
+			// "cdn.example.com:/photos/a%3Ab.jpg::800:600:webp:1704067200:85:cover"
+			wantSignature: "WnEpZ7bQp1Zvw8WDatlj2vu-4wYKWE3BFBizV8m__6k=",
+		},
+		{
+			name:  "colon in query",
+			path:  testPath,
+			query: "t=1:2",
+			// Signed data:
+			// "cdn.example.com:/photos/cat.jpg:t=1%3A2:800:600:webp:1704067200:85:cover"
+			wantSignature: "85kRoN8hhr4cCUkEUjNvUsvVYKjYYfWdeffx5TjJjqA=",
+		},
+		{
+			name:  "percent in path and query",
+			path:  "/photos/a%20b.jpg",
+			query: "q=50%",
+			// Signed data:
+			// "cdn.example.com:/photos/a%2520b.jpg:q=50%25:800:600:webp:1704067200:85:cover"
+			wantSignature: "MjDiiRdgXMrsjqPMsENtD4aG1hp1e-a6zy67vKTX4CQ=",
+		},
+	}
+
+	for _, tt := range vectors {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			req := &signature.Request{
+				SourceHost:  testHost,
+				SourcePath:  tt.path,
+				SourceQuery: tt.query,
+				Width:       800,
+				Height:      600,
+				Format:      testFormatWebP,
+				Quality:     85,
+				FitMode:     testFitCover,
+				Expires:     time.Unix(goldenExpiresUnix, 0),
+			}
+
+			gotSignature := signer.Sign(req)
+			t.Logf("Sign() of path %q query %q = %s", tt.path, tt.query, gotSignature)
+
+			if gotSignature != tt.wantSignature {
+				t.Errorf("Sign() = %q, want %q", gotSignature, tt.wantSignature)
+			}
+		})
+	}
+}

internal/signature/signature.go

@@ -10,6 +10,7 @@ import (
 	"fmt"
 	"net/url"
 	"strconv"
+	"strings"
 	"time"
 )
 
@@ -154,15 +155,17 @@ func (s *Signer) GenerateSignedURL(
 
 // buildSignatureData creates the string to be signed.
 // Format: "host:path:query:width:height:format:expiration:quality:fit"
-// All components are used verbatim (exact match). No normalization,
-// suffix matching, or wildcard expansion is performed. Quality and fit
-// are the effective transform values, so replaying a signed URL with a
-// different quality or fit mode fails verification.
+// The path and query go through escapePercentAndColon, so that a ":" in
+// either cannot be taken for the one between them; every other component
+// is used verbatim. No normalization, suffix matching, or wildcard
+// expansion is performed. Quality and fit are the effective transform
+// values, so replaying a signed URL with a different quality or fit mode
+// fails verification.
 func (s *Signer) buildSignatureData(req *Request) string {
 	return fmt.Sprintf("%s:%s:%s:%d:%d:%s:%d:%d:%s",
 		req.SourceHost,
-		req.SourcePath,
-		req.SourceQuery,
+		escapePercentAndColon(req.SourcePath),
+		escapePercentAndColon(req.SourceQuery),
 		req.Width,
 		req.Height,
 		req.Format,
@@ -172,6 +175,14 @@ func (s *Signer) buildSignatureData(req *Request) string {
 	)
 }
 
+// escapePercentAndColon returns s with each "%" written as "%25" and each
+// ":" as "%3A", so that the result holds no ":".
+func escapePercentAndColon(s string) string {
+	s = strings.ReplaceAll(s, "%", "%25")
+
+	return strings.ReplaceAll(s, ":", "%3A")
+}
+
 // ParseParams extracts signature and expiration from query parameters.
 func ParseParams(sig, expStr string) (string, time.Time, error) {
 	if expStr == "" {

script/loadtest

@@ -94,7 +94,8 @@ main() {
 }
 
 # start_containers starts a new origin and a new pixad, and waits up to 30
-# seconds for pixad's health check to pass.
+# seconds for pixad's health check to pass. pixad sees all of vegeta's
+# requests come from one address, so its image request limit is off.
 start_containers() {
     docker run -d --name "$run-origin" \
         --network "$run" --network-alias origin \
@@ -104,6 +105,7 @@ start_containers() {
         --network "$run" -p 127.0.0.1::8080 --health-interval=1s \
         -e PIXA_SIGNING_KEY="$(head -c 32 /dev/urandom | base64)" \
         -e PIXA_ALLOWLIST_HOSTS=origin -e PIXA_ALLOW_HTTP=true \
+        -e PIXA_IMAGE_REQUESTS_PER_MINUTE=0 \
         "$image" >/dev/null
 
     waited=0