client writer: handle pause before deocding

Adds a "cw-pause" client writer in the PROTOCOL phase that buffers
output when the client paused the transfer. This prevents content
decoding from blowing the buffer in the "cw-out" writer.

Added test_02_35 that downloads 2 100MB gzip bombs in parallel and
pauses after 1MB of decoded 0's.

This is a solution to issue #16280, with some limitations:
- cw-out still needs buffering of its own, since it can be paused
  "in the middle" of a write that started with some KB of gzipped
  zeros and exploded into several MB of calls to cw-out.
- cw-pause will then start buffering on its own *after* the write
  that caused the pause. cw-pause has no buffer limits, but the
  data it buffers is still content-encoded.
  Protocols like http/1.1 stop receiving, h2/h3 have window sizes,
  so the cw-pause buffer should not grow out of control, at least
  for these protocols.
- the current limit on cw-out's buffer is ~75MB (for whatever
  historical reason). A potential content-encoding that blows 16KB
  (the common h2 chunk size) into > 75MB would still blow the buffer,
  making the transfer fail. A gzip of 0's makes 16KB into ~16MB, so
  that still works.

A better solution would be to allow CURLE_AGAIN handling in the client
writer chain and make all content encoders handle that. This would stop
explosion of encoding on a pause right away. But this is a large change
of the deocoder operations.

Reported-by: lf- on github
Fixes #16280
Closes #16296
This commit is contained in:
Stefan Eissing 2025-02-10 17:40:11 +01:00 committed by Daniel Stenberg
parent 279a4772ae
commit f78700814d
No known key found for this signature in database
GPG Key ID: 5CC908FDB71E12C2
14 changed files with 468 additions and 55 deletions

View File

@ -144,6 +144,7 @@ LIB_CFILES = \
curl_threads.c \
curl_trc.c \
cw-out.c \
cw-pause.c \
dict.c \
dllmain.c \
doh.c \
@ -291,6 +292,7 @@ LIB_HFILES = \
curl_trc.h \
curlx.h \
cw-out.h \
cw-pause.h \
dict.h \
doh.h \
dynbuf.h \

View File

@ -32,6 +32,7 @@
#include "multiif.h"
#include "sendf.h"
#include "cw-out.h"
#include "cw-pause.h"
/* The last 3 #include files should be in this order */
#include "curl_printf.h"
@ -198,7 +199,7 @@ static CURLcode cw_out_ptr_flush(struct cw_out_ctx *ctx,
const char *buf, size_t blen,
size_t *pconsumed)
{
curl_write_callback wcb;
curl_write_callback wcb = NULL;
void *wcb_data;
size_t max_write, min_write;
size_t wlen, nwritten;
@ -222,7 +223,7 @@ static CURLcode cw_out_ptr_flush(struct cw_out_ctx *ctx,
Curl_set_in_callback(data, TRUE);
nwritten = wcb((char *)buf, 1, wlen, wcb_data);
Curl_set_in_callback(data, FALSE);
CURL_TRC_WRITE(data, "cw_out, wrote %zu %s bytes -> %zu",
CURL_TRC_WRITE(data, "[OUT] wrote %zu %s bytes -> %zu",
wlen, (otype == CW_OUT_BODY) ? "body" : "header",
nwritten);
if(CURL_WRITEFUNC_PAUSE == nwritten) {
@ -236,7 +237,7 @@ static CURLcode cw_out_ptr_flush(struct cw_out_ctx *ctx,
/* mark the connection as RECV paused */
data->req.keepon |= KEEP_RECV_PAUSE;
ctx->paused = TRUE;
CURL_TRC_WRITE(data, "cw_out, PAUSE requested by client");
CURL_TRC_WRITE(data, "[OUT] PAUSE requested by client");
break;
}
else if(CURL_WRITEFUNC_ERROR == nwritten) {
@ -326,11 +327,16 @@ static CURLcode cw_out_flush_chain(struct cw_out_ctx *ctx,
}
static CURLcode cw_out_append(struct cw_out_ctx *ctx,
struct Curl_easy *data,
cw_out_type otype,
const char *buf, size_t blen)
{
if(cw_out_bufs_len(ctx) + blen > DYN_PAUSE_BUFFER)
CURL_TRC_WRITE(data, "[OUT] paused, buffering %zu more bytes (%zu/%d)",
blen, cw_out_bufs_len(ctx), DYN_PAUSE_BUFFER);
if(cw_out_bufs_len(ctx) + blen > DYN_PAUSE_BUFFER) {
failf(data, "pause buffer not large enough -> CURLE_TOO_LARGE");
return CURLE_TOO_LARGE;
}
/* if we do not have a buffer, or it is of another type, make a new one.
* And for CW_OUT_HDS always make a new one, so we "replay" headers
@ -364,7 +370,7 @@ static CURLcode cw_out_do_write(struct cw_out_ctx *ctx,
if(ctx->buf) {
/* still have buffered data, append and flush */
result = cw_out_append(ctx, otype, buf, blen);
result = cw_out_append(ctx, data, otype, buf, blen);
if(result)
return result;
result = cw_out_flush_chain(ctx, data, &ctx->buf, flush_all);
@ -380,7 +386,8 @@ static CURLcode cw_out_do_write(struct cw_out_ctx *ctx,
return result;
if(consumed < blen) {
/* did not write all, append the rest */
result = cw_out_append(ctx, otype, buf + consumed, blen - consumed);
result = cw_out_append(ctx, data, otype,
buf + consumed, blen - consumed);
if(result)
goto out;
}
@ -430,23 +437,18 @@ bool Curl_cw_out_is_paused(struct Curl_easy *data)
return FALSE;
ctx = (struct cw_out_ctx *)cw_out;
CURL_TRC_WRITE(data, "cw-out is%spaused", ctx->paused ? "" : " not");
return ctx->paused;
}
static CURLcode cw_out_flush(struct Curl_easy *data,
bool unpause, bool flush_all)
struct Curl_cwriter *cw_out,
bool flush_all)
{
struct Curl_cwriter *cw_out;
struct cw_out_ctx *ctx = (struct cw_out_ctx *)cw_out;
CURLcode result = CURLE_OK;
cw_out = Curl_cwriter_get_by_type(data, &Curl_cwt_out);
if(cw_out) {
struct cw_out_ctx *ctx = (struct cw_out_ctx *)cw_out;
if(ctx->errored)
return CURLE_WRITE_ERROR;
if(unpause && ctx->paused)
ctx->paused = FALSE;
if(ctx->paused)
return CURLE_OK; /* not doing it */
@ -456,18 +458,37 @@ static CURLcode cw_out_flush(struct Curl_easy *data,
cw_out_bufs_free(ctx);
return result;
}
}
return result;
}
CURLcode Curl_cw_out_unpause(struct Curl_easy *data)
{
CURL_TRC_WRITE(data, "cw-out unpause");
return cw_out_flush(data, TRUE, FALSE);
struct Curl_cwriter *cw_out;
CURLcode result = CURLE_OK;
cw_out = Curl_cwriter_get_by_type(data, &Curl_cwt_out);
if(cw_out) {
struct cw_out_ctx *ctx = (struct cw_out_ctx *)cw_out;
CURL_TRC_WRITE(data, "[OUT] unpause");
ctx->paused = FALSE;
result = Curl_cw_pause_flush(data);
if(!result)
result = cw_out_flush(data, cw_out, FALSE);
}
return result;
}
CURLcode Curl_cw_out_done(struct Curl_easy *data)
{
CURL_TRC_WRITE(data, "cw-out done");
return cw_out_flush(data, FALSE, TRUE);
struct Curl_cwriter *cw_out;
CURLcode result = CURLE_OK;
cw_out = Curl_cwriter_get_by_type(data, &Curl_cwt_out);
if(cw_out) {
CURL_TRC_WRITE(data, "[OUT] done");
result = Curl_cw_pause_flush(data);
if(!result)
result = cw_out_flush(data, cw_out, TRUE);
}
return result;
}

242
lib/cw-pause.c Normal file
View File

@ -0,0 +1,242 @@
/***************************************************************************
* _ _ ____ _
* Project ___| | | | _ \| |
* / __| | | | |_) | |
* | (__| |_| | _ <| |___
* \___|\___/|_| \_\_____|
*
* Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
*
* This software is licensed as described in the file COPYING, which
* you should have received as part of this distribution. The terms
* are also available at https://curl.se/docs/copyright.html.
*
* You may opt to use, copy, modify, merge, publish, distribute and/or sell
* copies of the Software, and permit persons to whom the Software is
* furnished to do so, under the terms of the COPYING file.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
* SPDX-License-Identifier: curl
*
***************************************************************************/
#include "curl_setup.h"
#include <curl/curl.h>
#include "urldata.h"
#include "bufq.h"
#include "cfilters.h"
#include "headers.h"
#include "multiif.h"
#include "sendf.h"
#include "cw-pause.h"
/* The last 3 #include files should be in this order */
#include "curl_printf.h"
#include "curl_memory.h"
#include "memdebug.h"
/* body dynbuf sizes */
#define CW_PAUSE_BUF_CHUNK (16 * 1024)
/* when content decoding, write data in chunks */
#define CW_PAUSE_DEC_WRITE_CHUNK (4096)
struct cw_pause_buf {
struct cw_pause_buf *next;
struct bufq b;
int type;
};
static struct cw_pause_buf *cw_pause_buf_create(int type, size_t buflen)
{
struct cw_pause_buf *cwbuf = calloc(1, sizeof(*cwbuf));
if(cwbuf) {
cwbuf->type = type;
if(type & CLIENTWRITE_BODY)
Curl_bufq_init2(&cwbuf->b, CW_PAUSE_BUF_CHUNK, 1,
(BUFQ_OPT_SOFT_LIMIT|BUFQ_OPT_NO_SPARES));
else
Curl_bufq_init(&cwbuf->b, buflen, 1);
}
return cwbuf;
}
static void cw_pause_buf_free(struct cw_pause_buf *cwbuf)
{
if(cwbuf) {
Curl_bufq_free(&cwbuf->b);
free(cwbuf);
}
}
struct cw_pause_ctx {
struct Curl_cwriter super;
struct cw_pause_buf *buf;
size_t buf_total;
};
static CURLcode cw_pause_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
const char *buf, size_t nbytes);
static void cw_pause_close(struct Curl_easy *data,
struct Curl_cwriter *writer);
static CURLcode cw_pause_init(struct Curl_easy *data,
struct Curl_cwriter *writer);
struct Curl_cwtype Curl_cwt_pause = {
"cw-pause",
NULL,
cw_pause_init,
cw_pause_write,
cw_pause_close,
sizeof(struct cw_pause_ctx)
};
static CURLcode cw_pause_init(struct Curl_easy *data,
struct Curl_cwriter *writer)
{
struct cw_pause_ctx *ctx = writer->ctx;
(void)data;
ctx->buf = NULL;
return CURLE_OK;
}
static void cw_pause_bufs_free(struct cw_pause_ctx *ctx)
{
while(ctx->buf) {
struct cw_pause_buf *next = ctx->buf->next;
cw_pause_buf_free(ctx->buf);
ctx->buf = next;
}
}
static void cw_pause_close(struct Curl_easy *data, struct Curl_cwriter *writer)
{
struct cw_pause_ctx *ctx = writer->ctx;
(void)data;
cw_pause_bufs_free(ctx);
}
static CURLcode cw_pause_flush(struct Curl_easy *data,
struct Curl_cwriter *cw_pause)
{
struct cw_pause_ctx *ctx = (struct cw_pause_ctx *)cw_pause;
bool decoding = Curl_cwriter_is_content_decoding(data);
CURLcode result = CURLE_OK;
/* write the end of the chain until it blocks or gets empty */
while(ctx->buf && !Curl_cwriter_is_paused(data)) {
struct cw_pause_buf **plast = &ctx->buf;
size_t blen, wlen = 0;
const unsigned char *buf = NULL;
while((*plast)->next) /* got to last in list */
plast = &(*plast)->next;
if(Curl_bufq_peek(&(*plast)->b, &buf, &blen)) {
wlen = (decoding && ((*plast)->type & CLIENTWRITE_BODY)) ?
CURLMIN(blen, CW_PAUSE_DEC_WRITE_CHUNK) : blen;
result = Curl_cwriter_write(data, cw_pause->next, (*plast)->type,
(const char *)buf, wlen);
CURL_TRC_WRITE(data, "[PAUSE] flushed %zu/%zu bytes, type=%x -> %d",
wlen, ctx->buf_total, (*plast)->type, result);
Curl_bufq_skip(&(*plast)->b, wlen);
DEBUGASSERT(ctx->buf_total >= wlen);
ctx->buf_total -= wlen;
if(result)
return result;
}
else if((*plast)->type & CLIENTWRITE_EOS) {
result = Curl_cwriter_write(data, cw_pause->next, (*plast)->type,
(const char *)buf, 0);
CURL_TRC_WRITE(data, "[PAUSE] flushed 0/%zu bytes, type=%x -> %d",
ctx->buf_total, (*plast)->type, result);
}
if(Curl_bufq_is_empty(&(*plast)->b)) {
cw_pause_buf_free(*plast);
*plast = NULL;
}
}
return result;
}
static CURLcode cw_pause_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
const char *buf, size_t blen)
{
struct cw_pause_ctx *ctx = writer->ctx;
CURLcode result = CURLE_OK;
size_t wlen = 0;
bool decoding = Curl_cwriter_is_content_decoding(data);
if(ctx->buf && !Curl_cwriter_is_paused(data)) {
result = cw_pause_flush(data, writer);
if(result)
return result;
}
while(!ctx->buf && !Curl_cwriter_is_paused(data)) {
int wtype = type;
DEBUGASSERT(!ctx->buf);
/* content decoding might blow up size considerably, write smaller
* chunks to make pausing need buffer less. */
wlen = (decoding && (type & CLIENTWRITE_BODY)) ?
CURLMIN(blen, CW_PAUSE_DEC_WRITE_CHUNK) : blen;
if(wlen < blen)
wtype &= ~CLIENTWRITE_EOS;
result = Curl_cwriter_write(data, writer->next, wtype, buf, wlen);
CURL_TRC_WRITE(data, "[PAUSE] writing %zu/%zu bytes of type %x -> %d",
wlen, blen, wtype, result);
if(result)
return result;
buf += wlen;
blen -= wlen;
if(!blen)
return result;
}
do {
size_t nwritten = 0;
if(ctx->buf && (ctx->buf->type == type) && (type & CLIENTWRITE_BODY)) {
/* same type and body, append to current buffer which has a soft
* limit and should take everything up to OOM. */
result = Curl_bufq_cwrite(&ctx->buf->b, buf, blen, &nwritten);
}
else {
/* Need a new buf, type changed */
struct cw_pause_buf *cwbuf = cw_pause_buf_create(type, blen);
if(!cwbuf)
return CURLE_OUT_OF_MEMORY;
cwbuf->next = ctx->buf;
ctx->buf = cwbuf;
result = Curl_bufq_cwrite(&ctx->buf->b, buf, blen, &nwritten);
}
CURL_TRC_WRITE(data, "[PAUSE] buffer %zu more bytes of type %x, "
"total=%zu -> %d", nwritten, type, ctx->buf_total + wlen,
result);
if(result)
return result;
buf += nwritten;
blen -= nwritten;
ctx->buf_total += nwritten;
} while(blen);
return result;
}
CURLcode Curl_cw_pause_flush(struct Curl_easy *data)
{
struct Curl_cwriter *cw_pause;
CURLcode result = CURLE_OK;
cw_pause = Curl_cwriter_get_by_type(data, &Curl_cwt_pause);
if(cw_pause)
result = cw_pause_flush(data, cw_pause);
return result;
}

40
lib/cw-pause.h Normal file
View File

@ -0,0 +1,40 @@
#ifndef HEADER_CURL_CW_PAUSE_H
#define HEADER_CURL_CW_PAUSE_H
/***************************************************************************
* _ _ ____ _
* Project ___| | | | _ \| |
* / __| | | | |_) | |
* | (__| |_| | _ <| |___
* \___|\___/|_| \_\_____|
*
* Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
*
* This software is licensed as described in the file COPYING, which
* you should have received as part of this distribution. The terms
* are also available at https://curl.se/docs/copyright.html.
*
* You may opt to use, copy, modify, merge, publish, distribute and/or sell
* copies of the Software, and permit persons to whom the Software is
* furnished to do so, under the terms of the COPYING file.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
* SPDX-License-Identifier: curl
*
***************************************************************************/
#include "curl_setup.h"
#include "sendf.h"
/**
* The client writer type "cw-pause" that buffers writes for
* paused transfer writes.
*/
extern struct Curl_cwtype Curl_cwt_pause;
CURLcode Curl_cw_pause_flush(struct Curl_easy *data);
#endif /* HEADER_CURL_CW_PAUSE_H */

View File

@ -219,6 +219,7 @@ struct h2_stream_ctx {
BIT(bodystarted);
BIT(body_eos); /* the complete body has been added to `sendbuf` and
* is being/has been processed from there. */
BIT(write_paused); /* stream write is paused */
};
#define H2_STREAM_CTX(ctx,data) ((struct h2_stream_ctx *)(\
@ -289,14 +290,14 @@ static int32_t cf_h2_get_desired_local_win(struct Curl_cfilter *cf,
static CURLcode cf_h2_update_local_win(struct Curl_cfilter *cf,
struct Curl_easy *data,
struct h2_stream_ctx *stream,
bool paused)
struct h2_stream_ctx *stream)
{
struct cf_h2_ctx *ctx = cf->ctx;
int32_t dwsize;
int rv;
dwsize = paused ? 0 : cf_h2_get_desired_local_win(cf, data);
dwsize = (stream->write_paused || stream->xfer_result) ?
0 : cf_h2_get_desired_local_win(cf, data);
if(dwsize != stream->local_window_size) {
int32_t wsize = nghttp2_session_get_stream_effective_local_window_size(
ctx->h2, stream->id);
@ -332,13 +333,11 @@ static CURLcode cf_h2_update_local_win(struct Curl_cfilter *cf,
static CURLcode cf_h2_update_local_win(struct Curl_cfilter *cf,
struct Curl_easy *data,
struct h2_stream_ctx *stream,
bool paused)
struct h2_stream_ctx *stream)
{
(void)cf;
(void)data;
(void)stream;
(void)paused;
return CURLE_OK;
}
#endif /* !NGHTTP2_HAS_SET_LOCAL_WINDOW_SIZE */
@ -1058,7 +1057,7 @@ static void h2_xfer_write_resp_hd(struct Curl_cfilter *cf,
if(!stream->xfer_result) {
stream->xfer_result = Curl_xfer_write_resp_hd(data, buf, blen, eos);
if(!stream->xfer_result && !eos)
stream->xfer_result = cf_h2_update_local_win(cf, data, stream, FALSE);
stream->xfer_result = cf_h2_update_local_win(cf, data, stream);
if(stream->xfer_result)
CURL_TRC_CF(data, cf, "[%d] error %d writing %zu bytes of headers",
stream->id, stream->xfer_result, blen);
@ -1074,8 +1073,6 @@ static void h2_xfer_write_resp(struct Curl_cfilter *cf,
/* If we already encountered an error, skip further writes */
if(!stream->xfer_result)
stream->xfer_result = Curl_xfer_write_resp(data, buf, blen, eos);
if(!stream->xfer_result && !eos)
stream->xfer_result = cf_h2_update_local_win(cf, data, stream, FALSE);
/* If the transfer write is errored, we do not want any more data */
if(stream->xfer_result) {
struct cf_h2_ctx *ctx = cf->ctx;
@ -1085,6 +1082,17 @@ static void h2_xfer_write_resp(struct Curl_cfilter *cf,
nghttp2_submit_rst_stream(ctx->h2, 0, stream->id,
(uint32_t)NGHTTP2_ERR_CALLBACK_FAILURE);
}
else if(!stream->write_paused && Curl_xfer_write_is_paused(data)) {
CURL_TRC_CF(data, cf, "[%d] stream output paused", stream->id);
stream->write_paused = TRUE;
}
else if(stream->write_paused && !Curl_xfer_write_is_paused(data)) {
CURL_TRC_CF(data, cf, "[%d] stream output unpaused", stream->id);
stream->write_paused = FALSE;
}
if(!stream->xfer_result && !eos)
stream->xfer_result = cf_h2_update_local_win(cf, data, stream);
}
static CURLcode on_stream_frame(struct Curl_cfilter *cf,
@ -2579,7 +2587,10 @@ static CURLcode http2_data_pause(struct Curl_cfilter *cf,
DEBUGASSERT(data);
if(ctx && ctx->h2 && stream) {
CURLcode result = cf_h2_update_local_win(cf, data, stream, pause);
CURLcode result;
stream->write_paused = pause;
result = cf_h2_update_local_win(cf, data, stream);
if(result)
return result;

View File

@ -42,6 +42,7 @@
#include "connect.h"
#include "content_encoding.h"
#include "cw-out.h"
#include "cw-pause.h"
#include "vtls/vtls.h"
#include "vssh/ssh.h"
#include "easyif.h"
@ -433,21 +434,37 @@ static CURLcode do_init_writer_stack(struct Curl_easy *data)
if(result)
return result;
result = Curl_cwriter_create(&writer, data, &cw_download, CURL_CW_PROTOCOL);
if(result)
return result;
/* This places the "pause" writer behind the "download" writer that
* is added below. Meaning the "download" can do checks on content length
* and other things *before* write outs are buffered for paused transfers. */
result = Curl_cwriter_create(&writer, data, &Curl_cwt_pause,
CURL_CW_PROTOCOL);
if(!result) {
result = Curl_cwriter_add(data, writer);
if(result) {
if(result)
Curl_cwriter_free(data, writer);
}
if(result)
return result;
result = Curl_cwriter_create(&writer, data, &cw_download, CURL_CW_PROTOCOL);
if(!result) {
result = Curl_cwriter_add(data, writer);
if(result)
Curl_cwriter_free(data, writer);
}
if(result)
return result;
result = Curl_cwriter_create(&writer, data, &cw_raw, CURL_CW_RAW);
if(result)
return result;
if(!result) {
result = Curl_cwriter_add(data, writer);
if(result) {
if(result)
Curl_cwriter_free(data, writer);
}
if(result)
return result;
return result;
}
@ -494,6 +511,16 @@ struct Curl_cwriter *Curl_cwriter_get_by_type(struct Curl_easy *data,
return NULL;
}
bool Curl_cwriter_is_content_decoding(struct Curl_easy *data)
{
struct Curl_cwriter *writer;
for(writer = data->req.writer_stack; writer; writer = writer->next) {
if(writer->phase == CURL_CW_CONTENT_DECODE)
return TRUE;
}
return FALSE;
}
bool Curl_cwriter_is_paused(struct Curl_easy *data)
{
return Curl_cw_out_is_paused(data);

View File

@ -182,6 +182,8 @@ CURLcode Curl_cwriter_write(struct Curl_easy *data,
*/
bool Curl_cwriter_is_paused(struct Curl_easy *data);
bool Curl_cwriter_is_content_decoding(struct Curl_easy *data);
/**
* Unpause client writer and flush any buffered date to the client.
*/

View File

@ -880,6 +880,11 @@ CURLcode Curl_xfer_write_resp(struct Curl_easy *data,
return result;
}
bool Curl_xfer_write_is_paused(struct Curl_easy *data)
{
return Curl_cwriter_is_paused(data);
}
CURLcode Curl_xfer_write_resp_hd(struct Curl_easy *data,
const char *hd0, size_t hdlen, bool is_eos)
{

View File

@ -55,6 +55,8 @@ CURLcode Curl_xfer_write_resp(struct Curl_easy *data,
const char *buf, size_t blen,
bool is_eos);
bool Curl_xfer_write_is_paused(struct Curl_easy *data);
/**
* Write a single "header" line from a server response.
* @param hd0 the 0-terminated, single header line

View File

@ -159,6 +159,7 @@ struct transfer {
int paused;
int resumed;
int done;
CURLcode result;
};
static size_t transfer_count = 1;
@ -240,6 +241,7 @@ static int setup(CURL *hnd, const char *url, struct transfer *t,
curl_easy_setopt(hnd, CURLOPT_HTTP_VERSION, http_version);
curl_easy_setopt(hnd, CURLOPT_SSL_VERIFYPEER, 0L);
curl_easy_setopt(hnd, CURLOPT_SSL_VERIFYHOST, 0L);
curl_easy_setopt(hnd, CURLOPT_ACCEPT_ENCODING, "");
curl_easy_setopt(hnd, CURLOPT_BUFFERSIZE, (long)(128 * 1024));
curl_easy_setopt(hnd, CURLOPT_WRITEFUNCTION, my_write_cb);
curl_easy_setopt(hnd, CURLOPT_WRITEDATA, t);
@ -472,7 +474,9 @@ int main(int argc, char *argv[])
t = get_transfer_for_easy(e);
if(t) {
t->done = 1;
fprintf(stderr, "[t-%d] FINISHED\n", t->idx);
t->result = m->data.result;
fprintf(stderr, "[t-%d] FINISHED with result %d\n",
t->idx, t->result);
if(use_earlydata) {
curl_off_t sent;
curl_easy_getinfo(e, CURLINFO_EARLYDATA_SENT_T, &sent);
@ -551,6 +555,8 @@ int main(int argc, char *argv[])
curl_easy_cleanup(t->easy);
t->easy = NULL;
}
if(t->result)
result = t->result;
}
free(transfers);

View File

@ -56,6 +56,7 @@ class TestDownload:
env.make_data_file(indir=indir, fname="data-1m", fsize=1024*1024)
env.make_data_file(indir=indir, fname="data-10m", fsize=10*1024*1024)
env.make_data_file(indir=indir, fname="data-50m", fsize=50*1024*1024)
env.make_data_gzipbomb(indir=indir, fname="bomb-100m.txt", fsize=100*1024*1024)
# download 1 file
@pytest.mark.parametrize("proto", ['http/1.1', 'h2', 'h3'])
@ -405,7 +406,7 @@ class TestDownload:
'-n', f'{count}', '-m', f'{max_parallel}', '-a',
'-A', f'{abort_offset}', '-V', proto, url
])
r.check_exit_code(0)
r.check_exit_code(42) # CURLE_ABORTED_BY_CALLBACK
srcfile = os.path.join(httpd.docs_dir, docname)
# downloads should be there, but not necessarily complete
self.check_downloads(client, srcfile, count, complete=False)
@ -434,7 +435,7 @@ class TestDownload:
'-n', f'{count}', '-m', f'{max_parallel}', '-a',
'-F', f'{fail_offset}', '-V', proto, url
])
r.check_exit_code(0)
r.check_exit_code(23) # CURLE_WRITE_ERROR
srcfile = os.path.join(httpd.docs_dir, docname)
# downloads should be there, but not necessarily complete
self.check_downloads(client, srcfile, count, complete=False)
@ -615,11 +616,11 @@ class TestDownload:
assert reused_session, 'session was not reused for 2nd transfer'
assert earlydata[0] == 0, f'{earlydata}'
if proto == 'http/1.1':
assert earlydata[1] == 69, f'{earlydata}'
assert earlydata[1] == 111, f'{earlydata}'
elif proto == 'h2':
assert earlydata[1] == 107, f'{earlydata}'
assert earlydata[1] == 127, f'{earlydata}'
elif proto == 'h3':
assert earlydata[1] == 67, f'{earlydata}'
assert earlydata[1] == 109, f'{earlydata}'
@pytest.mark.parametrize("proto", ['http/1.1', 'h2'])
@pytest.mark.parametrize("max_host_conns", [0, 1, 5])
@ -688,3 +689,30 @@ class TestDownload:
n = int(m.group(1))
assert n <= max_total_conns
assert matched_lines > 0
# 2 parallel transers, pause and resume. Load a 100 MB zip bomb from
# the server with "Content-Encoding: gzip" that gets exloded during
# response writing to the client. Client pauses after 1MB unzipped data
# and causes buffers to fill while the server sends more response
# data.
# * http/1.1: not much buffering is done as curl does no longer
# serve the connections that are paused
# * h2/h3: server continues sending what the stream window allows and
# since the one connection involved unpaused transfers, data continues
# to be received, requiring buffering.
@pytest.mark.parametrize("proto", ['http/1.1', 'h2', 'h3'])
def test_02_35_pause_bomb(self, env: Env, httpd, nghttpx, proto):
if proto == 'h3' and not env.have_h3():
pytest.skip("h3 not supported")
count = 2
pause_offset = 1024 * 1024
docname = 'bomb-100m.txt.var'
url = f'https://localhost:{env.https_port}/{docname}'
client = LocalClient(name='hx-download', env=env)
if not client.exists():
pytest.skip(f'example client not built: {client.name}')
r = client.run(args=[
'-n', f'{count}', '-m', f'{count}',
'-P', f'{pause_offset}', '-V', proto, url
])
r.check_exit_code(0)

View File

@ -234,7 +234,7 @@ class TestCaddy:
earlydata[int(m.group(1))] = int(m.group(2))
assert earlydata[0] == 0, f'{earlydata}'
if proto == 'h3':
assert earlydata[1] == 71, f'{earlydata}'
assert earlydata[1] == 113, f'{earlydata}'
else:
# Caddy does not support early data on TCP
assert earlydata[1] == 0, f'{earlydata}'

View File

@ -24,6 +24,7 @@
#
###########################################################################
#
import gzip
import logging
import os
import re
@ -618,3 +619,24 @@ class Env:
i = int(fsize / line_length) + 1
fd.write(f"{i:09d}-{s}"[0:remain-1] + "\n")
return fpath
def make_data_gzipbomb(self, indir: str, fname: str, fsize: int) -> str:
fpath = os.path.join(indir, fname)
gzpath = f'{fpath}.gz'
varpath = f'{fpath}.var'
with open(fpath, 'w') as fd:
fd.write('not what we are looking for!\n')
count = int(fsize / 1024)
zero1k = bytearray(1024)
with gzip.open(gzpath, 'wb') as fd:
for _ in range(count):
fd.write(zero1k)
with open(varpath, 'w') as fd:
fd.write(f'URI: {fname}\n')
fd.write('\n')
fd.write(f'URI: {fname}.gz\n')
fd.write('Content-Type: text/plain\n')
fd.write('Content-Encoding: x-gzip\n')
fd.write('\n')
return fpath

View File

@ -48,7 +48,7 @@ class Httpd:
'authn_core', 'authn_file',
'authz_user', 'authz_core', 'authz_host',
'auth_basic', 'auth_digest',
'alias', 'env', 'filter', 'headers', 'mime', 'setenvif',
'alias', 'env', 'filter', 'headers', 'mime', 'setenvif', 'negotiation',
'socache_shmcb',
'rewrite', 'http2', 'ssl', 'proxy', 'proxy_http', 'proxy_connect',
'brotli',
@ -269,6 +269,8 @@ class Httpd:
f'Listen {self.env.proxys_port}',
f'TypesConfig "{self._conf_dir}/mime.types',
'SSLSessionCache "shmcb:ssl_gcache_data(32000)"',
'AddEncoding x-gzip .gz .tgz .gzip',
'AddHandler type-map .var',
]
if 'base' in self._extra_configs:
conf.extend(self._extra_configs['base'])
@ -399,8 +401,11 @@ class Httpd:
fd.write("\n".join(conf))
with open(os.path.join(self._conf_dir, 'mime.types'), 'w') as fd:
fd.write("\n".join([
'text/plain txt',
'text/html html',
'application/json json',
'application/x-gzip gzip',
'application/x-gzip gz',
''
]))