1.8.2: intermittent HTTP response corruption on keep-alive connections (regression from the CVE-2026-87776 fix)
Valutazione
- Difficoltà
- 5/5
- Tempo stimato
- Più di una settimana
- Idoneità per principianti
- 8/100
- Tipo di issue
- Bug
- Chiarezza
- Abbastanza chiara
- Stato di attività
- Attiva
- Stack tecnologico
- express, javascript, nodejs
- Ambito
- backend, networking
Direzione di ricerca
The suspected cause is the close listener that 1.8.2 adds to destroy the zlib stream and fall back to the raw response, so start with the diff between 1.8.1 and 1.8.2 in the compression source. There is no minimal repro yet, so the first useful step is building one that combines streamed bodies, client aborts and a keep-alive agent. Done means that repro fails on 1.8.2 and passes on 1.8.1, with the race fixed and the framing verified.
Scritto dal modello di indicizzazione a partire dal testo della issue.
Descrizione
Summary
After upgrading compression 1.8.1 → 1.8.2 (the fix for the client-abort zlib memory leak, CVE-2026-87776), our Express 4 API's integration test suite became reliably flaky: roughly 1 request per ~600 intermittently gets a corrupted or misframed response on keep-alive connections. Pinning back to 1.8.1 — with every other dependency unchanged — makes the suite stable again.
Symptoms observed (varies per occurrence)
Error: Parse Error: Expected HTTP/, RTSP/ or ICE/raised by the Node HTTP client — the bytes on the socket do not start with a status line, i.e. the previous response's framing didn't end where its headers said it would;- responses seemingly answered to the wrong request on the socket (a route that must 401 returns 404 or 400, a GET that must 200 returns 404);
- empty bodies (
{}after JSON parsing) on responses that are otherwise 200; socket hang upmid-body.
The failing requests are ordinary JSON endpoints; the suite also exercises streamed responses (image resizing piped through sharp) and requests aborted client-side, which run before/concurrently on the same keep-alive agent.
Evidence (A/B, single variable)
Same app, same Node, same test suite, ~600 HTTP requests per run:
| compression | runs | failing runs |
|---|---|---|
| 1.8.1 | 6+ | 0 |
| 1.8.2 | 3 | 3 (1 corrupted response each) |
Every other package was held constant during the comparison (we bisected it out of a larger dependency upgrade; morgan and body-parser bumps were cleared separately).
Suspected mechanism
1.8.2 adds a close listener that destroys the zlib stream when the response closes, and falls back to the raw response if the stream is already gone. If that teardown races a response that is still being written on a kept-alive socket (or runs on close of a completed response, since close also fires after normal end), part of the compressed body can be dropped or written raw, so the bytes on the wire no longer match the declared framing — and every subsequent response on that socket is misparsed.
Environment
- compression 1.8.2 / express 4.22.1 / Node v24.19.0 (macOS, also reproduced with express 4.21.2's stack)
- Client: node core
httpwithkeepAlive: trueagents (mocha + chai-http/superagent)
I don't have a minimal standalone repro yet — a naive script (keep-alive agent + aborted compressed responses) didn't trigger it, so the race window seems to need the larger app's timing (streamed bodies + aborts + concurrent sockets). Happy to run instrumented builds or provide more detail from our suite if useful.
- Lingua principale
- JavaScript
- Stelle
- 2.8k
- Fork
- 260
- Metriche di merge delle PR
- Nessuna PR unita negli ultimi 30g
Preparare l'ambiente
- Nessun Dockerfile né file Docker Compose
- Nessun modello di pull request
- Leggi la guida per i contributori
Come iniziare
- Leggi tutta la issue e poi la guida ai contributi del progetto.
- Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
- Fai un fork del repository e lavora su un branch.
- Apri una pull request che faccia riferimento al numero della issue.
Altre issue di expressjs/compression
-
Compression middleware changes behavior of undefined statusText in writeHeadForse già presa @dyk1454683243-sudo l’ha presa 29 giorni fa. Apertabug
Difficoltà 3/5 1-2 giorni Idoneità per principianti 62/100
expressjs/compression#254 · 2 commenti ·
-
Compression v2Forse di nuovo libera @UlisesGascon l’ha presa 522 giorni fa e non c’è nessuna pull request aperta. Aperta
expressjs/compression#234 · 1 reazione · 2 assegnatari ·
-
awaiting more info bug
Difficoltà 4/5 3-5 giorni Idoneità per principianti 35/100
expressjs/compression#220 · 11 commenti · 2 reazioni ·
-
Support zstdForse già presa @polys l’ha presa 433 giorni fa. Apertaenhancement future
Difficoltà 4/5 3-5 giorni Idoneità per principianti 30/100
expressjs/compression#217 · 4 commenti · 12 reazioni ·
-
Express returns a non-compliant HTTP/206 response when gzip is enabledForse già presa @vibhor-aggr l’ha presa 130 giorni fa. Aperta
Difficoltà 4/5 3-5 giorni Idoneità per principianti 35/100
expressjs/compression#185 · 9 commenti · 1 reazione ·
Tutte le issue di expressjs/compression
Issue simili
-
[APP/MODULE]Apertaenhancement
Difficoltà 2/5 1-3 ore Idoneità per principianti 68/100
awesome-android-root/awesome-android-root#195 ·
I maintainer di solito rispondono entro 1 giorno
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 72/100
subhranshudash13-dotcom/Contribo#74 ·
I maintainer di solito rispondono entro 1 giorno
-
enhancement
Difficoltà 2/5 1-3 ore Idoneità per principianti 62/100
I maintainer di solito rispondono entro 1 giorno
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 84/100
qq15725/modern-screenshot#174 ·
-
Difficoltà 1/5 Meno di un'ora Idoneità per principianti 67/100
ThatGuySam/doesitarm#1114 ·