curl/lib/cw-out.c
Viktor Szakats 193cb00ce9
build: stop overriding standard memory allocation functions
Before this patch curl used the C preprocessor to override standard
memory allocation symbols: malloc, calloc, strdup, realloc, free.
The goal of these is to replace them with curl's debug wrappers in
`CURLDEBUG` builds, another was to replace them with the wrappers
calling user-defined allocators in libcurl. This solution needed a bunch
of workarounds to avoid breaking external headers: it relied on include
order to do the overriding last. For "unity" builds it needed to reset
overrides before external includes. Also in test apps, which are always
built as single source files. It also needed the `(symbol)` trick
to avoid overrides in some places. This would still not fix cases where
the standard symbols were macros. It was also fragile and difficult
to figure out which was the actual function behind an alloc or free call
in a specific piece of code. This in turn caused bugs where the wrong
allocator was accidentally called.

To avoid these problems, this patch replaces this solution with
`curlx_`-prefixed allocator macros, and mapping them _once_ to either
the libcurl wrappers, the debug wrappers or the standard ones, matching
the rest of the code in libtests.

This concludes the long journey to avoid redefining standard functions
in the curl codebase.

Note: I did not update `packages/OS400/*.c` sources. They did not
`#include` `curl_setup.h`, `curl_memory.h` or `memdebug.h`, meaning
the overrides were never applied to them. This may or may not have been
correct. For now I suppressed the direct use of standard allocators
via a local `.checksrc`. Probably they (except for `curlcl.c`) should be
updated to include `curl_setup.h` and use the `curlx_` macros.

This patch changes mappings in two places:
- `lib/curl_threads.c` in libtests: Before this patch it mapped to
  libcurl allocators. After, it maps to standard allocators, like
  the rest of libtests code.
- `units`: before this patch it mapped to standard allocators. After, it
  maps to libcurl allocators.

Also:
- drop all position-dependent `curl_memory.h` and `memdebug.h` includes,
  and delete the now unnecessary headers.
- rename `Curl_tcsdup` macro to `curlx_tcsdup` and define like the other
  allocators.
- map `curlx_strdup()` to `_strdup()` on Windows (was: `strdup()`).
  To fix warnings silenced via `_CRT_NONSTDC_NO_DEPRECATE`.
- multibyte: map `curlx_convert_*()` to `_strdup()` on Windows
  (was: `strdup()`).
- src: do not reuse the `strdup` name for the local replacement.
- lib509: call `_strdup()` on Windows (was: `strdup()`).
- test1132: delete test obsoleted by this patch.
- CHECKSRC.md: update text for `SNPRINTF`.
- checksrc: ban standard allocator symbols.

Follow-up to b12da22db1 #18866
Follow-up to db98daab05 #18844
Follow-up to 4deea9396b #18814
Follow-up to 9678ff5b1b #18776
Follow-up to 10bac43b87 #18774
Follow-up to 20142f5d06 #18634
Follow-up to bf7375ecc5 #18503
Follow-up to 9863599d69 #18502
Follow-up to 3bb5e58c10 #17827

Closes #19626
2025-11-28 10:44:26 +01:00

525 lines
15 KiB
C

/***************************************************************************
* _ _ ____ _
* 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 "cfilters.h"
#include "headers.h"
#include "multiif.h"
#include "sendf.h"
#include "transfer.h"
#include "cw-out.h"
#include "cw-pause.h"
/**
* OVERALL DESIGN of this client writer
*
* The 'cw-out' writer is supposed to be the last writer in a transfer's
* stack. It is always added when that stack is initialized. Its purpose
* is to pass BODY and HEADER bytes to the client-installed callback
* functions.
*
* These callback may return `CURL_WRITEFUNC_PAUSE` to indicate that the
* data had not been written and the whole transfer should stop receiving
* new data. Or at least, stop calling the functions. When the transfer
* is "unpaused" by the client, the previous data shall be passed as
* if nothing happened.
*
* The `cw-out` writer therefore manages buffers for bytes that could
* not be written. Data that was already in flight from the server also
* needs buffering on paused transfer when it arrives.
*
* In addition, the writer allows buffering of "small" body writes,
* so client functions are called less often. That is only enabled on a
* number of conditions.
*
* HEADER and BODY data may arrive in any order. For paused transfers,
* a list of `struct cw_out_buf` is kept for `cw_out_type` types. The
* list may be: [BODY]->[HEADER]->[BODY]->[HEADER]....
* When unpausing, this list is "played back" to the client callbacks.
*
* The amount of bytes being buffered is limited by `DYN_PAUSE_BUFFER`
* and when that is exceeded `CURLE_TOO_LARGE` is returned as error.
*/
typedef enum {
CW_OUT_NONE,
CW_OUT_BODY,
CW_OUT_BODY_0LEN,
CW_OUT_HDS
} cw_out_type;
struct cw_out_buf {
struct cw_out_buf *next;
struct dynbuf b;
cw_out_type type;
};
static struct cw_out_buf *cw_out_buf_create(cw_out_type otype)
{
struct cw_out_buf *cwbuf = curlx_calloc(1, sizeof(*cwbuf));
if(cwbuf) {
cwbuf->type = otype;
curlx_dyn_init(&cwbuf->b, DYN_PAUSE_BUFFER);
}
return cwbuf;
}
static void cw_out_buf_free(struct cw_out_buf *cwbuf)
{
if(cwbuf) {
curlx_dyn_free(&cwbuf->b);
curlx_free(cwbuf);
}
}
struct cw_out_ctx {
struct Curl_cwriter super;
struct cw_out_buf *buf;
BIT(paused);
BIT(errored);
};
static CURLcode cw_out_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
const char *buf, size_t nbytes);
static void cw_out_close(struct Curl_easy *data, struct Curl_cwriter *writer);
static CURLcode cw_out_init(struct Curl_easy *data,
struct Curl_cwriter *writer);
const struct Curl_cwtype Curl_cwt_out = {
"cw-out",
NULL,
cw_out_init,
cw_out_write,
cw_out_close,
sizeof(struct cw_out_ctx)
};
static CURLcode cw_out_init(struct Curl_easy *data,
struct Curl_cwriter *writer)
{
struct cw_out_ctx *ctx = writer->ctx;
(void)data;
ctx->buf = NULL;
return CURLE_OK;
}
static void cw_out_bufs_free(struct cw_out_ctx *ctx)
{
while(ctx->buf) {
struct cw_out_buf *next = ctx->buf->next;
cw_out_buf_free(ctx->buf);
ctx->buf = next;
}
}
static size_t cw_out_bufs_len(struct cw_out_ctx *ctx)
{
struct cw_out_buf *cwbuf = ctx->buf;
size_t len = 0;
while(cwbuf) {
len += curlx_dyn_len(&cwbuf->b);
cwbuf = cwbuf->next;
}
return len;
}
static void cw_out_close(struct Curl_easy *data, struct Curl_cwriter *writer)
{
struct cw_out_ctx *ctx = writer->ctx;
(void)data;
cw_out_bufs_free(ctx);
}
/**
* Return the current curl_write_callback and user_data for the buf type
*/
static void cw_get_writefunc(struct Curl_easy *data, cw_out_type otype,
curl_write_callback *pwcb, void **pwcb_data,
size_t *pmax_write, size_t *pmin_write)
{
switch(otype) {
case CW_OUT_BODY:
case CW_OUT_BODY_0LEN:
*pwcb = data->set.fwrite_func;
*pwcb_data = data->set.out;
*pmax_write = CURL_MAX_WRITE_SIZE;
/* if we ever want buffering of BODY output, we can set `min_write`
* the preferred size. The default should always be to pass data
* to the client as it comes without delay */
*pmin_write = 0;
break;
case CW_OUT_HDS:
*pwcb = data->set.fwrite_header ? data->set.fwrite_header :
(data->set.writeheader ? data->set.fwrite_func : NULL);
*pwcb_data = data->set.writeheader;
*pmax_write = 0; /* do not chunk-write headers, write them as they are */
*pmin_write = 0;
break;
default:
*pwcb = NULL;
*pwcb_data = NULL;
*pmax_write = CURL_MAX_WRITE_SIZE;
*pmin_write = 0;
}
}
static CURLcode cw_out_cb_write(struct cw_out_ctx *ctx,
struct Curl_easy *data,
curl_write_callback wcb,
void *wcb_data,
cw_out_type otype,
const char *buf, size_t blen,
size_t *pnwritten)
{
size_t nwritten;
CURLcode result;
DEBUGASSERT(data->conn);
*pnwritten = 0;
Curl_set_in_callback(data, TRUE);
nwritten = wcb((char *)CURL_UNCONST(buf), 1, blen, wcb_data);
Curl_set_in_callback(data, FALSE);
CURL_TRC_WRITE(data, "[OUT] wrote %zu %s bytes -> %zu",
blen, (otype == CW_OUT_HDS) ? "header" : "body",
nwritten);
if(CURL_WRITEFUNC_PAUSE == nwritten) {
if(data->conn->handler->flags & PROTOPT_NONETWORK) {
/* Protocols that work without network cannot be paused. This is
actually only FILE:// just now, and it cannot pause since the
transfer is not done using the "normal" procedure. */
failf(data, "Write callback asked for PAUSE when not supported");
return CURLE_WRITE_ERROR;
}
ctx->paused = TRUE;
CURL_TRC_WRITE(data, "[OUT] PAUSE requested by client");
result = Curl_xfer_pause_recv(data, TRUE);
return result ? result : CURLE_AGAIN;
}
else if(CURL_WRITEFUNC_ERROR == nwritten) {
failf(data, "client returned ERROR on write of %zu bytes", blen);
return CURLE_WRITE_ERROR;
}
else if(nwritten != blen) {
failf(data, "Failure writing output to destination, "
"passed %zu returned %zd", blen, nwritten);
return CURLE_WRITE_ERROR;
}
*pnwritten = nwritten;
return CURLE_OK;
}
static CURLcode cw_out_ptr_flush(struct cw_out_ctx *ctx,
struct Curl_easy *data,
cw_out_type otype,
bool flush_all,
const char *buf, size_t blen,
size_t *pconsumed)
{
curl_write_callback wcb = NULL;
void *wcb_data;
size_t max_write, min_write;
size_t wlen, nwritten;
CURLcode result;
/* If we errored once, we do not invoke the client callback again */
if(ctx->errored)
return CURLE_WRITE_ERROR;
/* write callbacks may get NULLed by the client between calls. */
cw_get_writefunc(data, otype, &wcb, &wcb_data, &max_write, &min_write);
if(!wcb) {
*pconsumed = blen;
return CURLE_OK;
}
*pconsumed = 0;
if(otype == CW_OUT_BODY_0LEN) {
DEBUGASSERT(!blen);
return cw_out_cb_write(ctx, data, wcb, wcb_data, otype,
buf, blen, &nwritten);
}
else {
while(blen && !ctx->paused) {
if(!flush_all && blen < min_write)
break;
wlen = max_write ? CURLMIN(blen, max_write) : blen;
result = cw_out_cb_write(ctx, data, wcb, wcb_data, otype,
buf, wlen, &nwritten);
if(result)
return result;
*pconsumed += nwritten;
blen -= nwritten;
buf += nwritten;
}
}
return CURLE_OK;
}
static CURLcode cw_out_buf_flush(struct cw_out_ctx *ctx,
struct Curl_easy *data,
struct cw_out_buf *cwbuf,
bool flush_all)
{
CURLcode result = CURLE_OK;
if(curlx_dyn_len(&cwbuf->b) || (cwbuf->type == CW_OUT_BODY_0LEN)) {
size_t consumed;
result = cw_out_ptr_flush(ctx, data, cwbuf->type, flush_all,
curlx_dyn_ptr(&cwbuf->b),
curlx_dyn_len(&cwbuf->b),
&consumed);
if(result && (result != CURLE_AGAIN))
return result;
result = CURLE_OK;
if(consumed) {
if(consumed == curlx_dyn_len(&cwbuf->b)) {
curlx_dyn_free(&cwbuf->b);
}
else {
DEBUGASSERT(consumed < curlx_dyn_len(&cwbuf->b));
result = curlx_dyn_tail(&cwbuf->b,
curlx_dyn_len(&cwbuf->b) - consumed);
if(result)
return result;
}
}
}
return result;
}
static CURLcode cw_out_flush_chain(struct cw_out_ctx *ctx,
struct Curl_easy *data,
struct cw_out_buf **pcwbuf,
bool flush_all)
{
struct cw_out_buf *cwbuf = *pcwbuf;
CURLcode result;
if(!cwbuf)
return CURLE_OK;
if(ctx->paused)
return CURLE_OK;
/* write the end of the chain until it blocks or gets empty */
while(cwbuf->next) {
struct cw_out_buf **plast = &cwbuf->next;
while((*plast)->next)
plast = &(*plast)->next;
result = cw_out_flush_chain(ctx, data, plast, flush_all);
if(result)
return result;
if(*plast) {
/* could not write last, paused again? */
DEBUGASSERT(ctx->paused);
return CURLE_OK;
}
}
result = cw_out_buf_flush(ctx, data, cwbuf, flush_all);
if(result)
return result;
if(!curlx_dyn_len(&cwbuf->b)) {
cw_out_buf_free(cwbuf);
*pcwbuf = NULL;
}
return CURLE_OK;
}
static CURLcode cw_out_append(struct cw_out_ctx *ctx,
struct Curl_easy *data,
cw_out_type otype,
const char *buf, size_t blen)
{
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
* exactly as they came in */
if(!ctx->buf || (ctx->buf->type != otype) || (otype == CW_OUT_HDS)) {
struct cw_out_buf *cwbuf = cw_out_buf_create(otype);
if(!cwbuf)
return CURLE_OUT_OF_MEMORY;
cwbuf->next = ctx->buf;
ctx->buf = cwbuf;
}
DEBUGASSERT(ctx->buf && (ctx->buf->type == otype));
return curlx_dyn_addn(&ctx->buf->b, buf, blen);
}
static CURLcode cw_out_do_write(struct cw_out_ctx *ctx,
struct Curl_easy *data,
cw_out_type otype,
bool flush_all,
const char *buf, size_t blen)
{
CURLcode result = CURLE_OK;
/* if we have buffered data and it is a different type than what
* we are writing now, try to flush all */
if(ctx->buf && ctx->buf->type != otype) {
result = cw_out_flush_chain(ctx, data, &ctx->buf, TRUE);
if(result)
goto out;
}
if(ctx->buf) {
/* still have buffered data, append and flush */
result = cw_out_append(ctx, data, otype, buf, blen);
if(result)
goto out;
result = cw_out_flush_chain(ctx, data, &ctx->buf, flush_all);
if(result)
goto out;
}
else {
/* nothing buffered, try direct write */
size_t consumed;
result = cw_out_ptr_flush(ctx, data, otype, flush_all,
buf, blen, &consumed);
if(result && (result != CURLE_AGAIN))
return result;
result = CURLE_OK;
if(consumed < blen) {
/* did not write all, append the rest */
result = cw_out_append(ctx, data, otype,
buf + consumed, blen - consumed);
if(result)
goto out;
}
}
out:
if(result) {
/* We do not want to invoked client callbacks a second time after
* encountering an error. See issue #13337 */
ctx->errored = TRUE;
cw_out_bufs_free(ctx);
}
return result;
}
static CURLcode cw_out_write(struct Curl_easy *data,
struct Curl_cwriter *writer, int type,
const char *buf, size_t blen)
{
struct cw_out_ctx *ctx = writer->ctx;
CURLcode result;
bool flush_all = !!(type & CLIENTWRITE_EOS);
if((type & CLIENTWRITE_BODY) ||
((type & CLIENTWRITE_HEADER) && data->set.include_header)) {
cw_out_type otype = (!blen && (type & CLIENTWRITE_0LEN)) ?
CW_OUT_BODY_0LEN : CW_OUT_BODY;
result = cw_out_do_write(ctx, data, otype, flush_all, buf, blen);
if(result)
return result;
}
if(type & (CLIENTWRITE_HEADER|CLIENTWRITE_INFO)) {
result = cw_out_do_write(ctx, data, CW_OUT_HDS, flush_all, buf, blen);
if(result)
return result;
}
return CURLE_OK;
}
bool Curl_cw_out_is_paused(struct Curl_easy *data)
{
struct Curl_cwriter *cw_out;
struct cw_out_ctx *ctx;
cw_out = Curl_cwriter_get_by_type(data, &Curl_cwt_out);
if(!cw_out)
return FALSE;
ctx = (struct cw_out_ctx *)cw_out;
return ctx->paused;
}
static CURLcode cw_out_flush(struct Curl_easy *data,
struct Curl_cwriter *cw_out,
bool flush_all)
{
struct cw_out_ctx *ctx = (struct cw_out_ctx *)cw_out;
CURLcode result = CURLE_OK;
if(ctx->errored)
return CURLE_WRITE_ERROR;
if(ctx->paused)
return CURLE_OK; /* not doing it */
result = cw_out_flush_chain(ctx, data, &ctx->buf, flush_all);
if(result) {
ctx->errored = TRUE;
cw_out_bufs_free(ctx);
return result;
}
return result;
}
CURLcode Curl_cw_out_unpause(struct Curl_easy *data)
{
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)
{
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;
}