Skip to content

http2: pre-allocate body buffer to eliminate realloc overhead #11972

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
yadoer wants to merge 1 commit into
base: master
Choose a base branch
from
+42 −1
Open

http2: pre-allocate body buffer to eliminate realloc overhead #11972

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
43 changes: 42 additions & 1 deletion src/http_server/flb_http_server_http2.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -743,7 +743,48 @@ static int http2_data_chunk_recv_callback(nghttp2_session *inner_session,
}

if (stream->request.body == NULL) {
stream->request.body = cfl_sds_create_size(len);
/*
* Pre-allocate body buffer to avoid repeated reallocs as 16KB HTTP/2
* frames arrive. Three strategies, in priority order:
* 1. Content-Length header — exact size (non-gRPC HTTP/2)
* 2. gRPC 5-byte frame header — parse message length for exact size
* 3. Fallback to frame size (len) — original behavior
*/
size_t initial_size;

@cosmo0920 cosmo0920 Jun 22, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We need to define variables on the top of functions due to coding conventions on this project.
See: https://github.com/fluent/fluent-bit/blob/master/CONTRIBUTING.md#variable-definitions


if (stream->request.content_length > 0) {
initial_size = stream->request.content_length;
}
else if (len >= 5 &&
stream->request.content_type != NULL &&
strncmp(stream->request.content_type, "application/grpc", 16) == 0) {
/* gRPC length-prefixed frame: 1 byte flags + 4 bytes message length */
uint32_t grpc_msg_len = ((uint32_t) data[1] << 24) |
((uint32_t) data[2] << 16) |
((uint32_t) data[3] << 8) |
((uint32_t) data[4]);
if ((size_t) grpc_msg_len > SIZE_MAX - 5) {
stream->status = HTTP_STREAM_STATUS_ERROR;
return -1;
}
initial_size = (size_t) grpc_msg_len + 5;

@chatgpt-codex-connector chatgpt-codex-connector Bot Jun 21, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Allocate at least the bytes already received

When a gRPC DATA callback contains more bytes than the first length-prefixed message advertises (for example a zero-length first message followed by another message in the same DATA frame, or a malformed gRPC request), this sets initial_size below len; the later cfl_sds_create_size(initial_size) followed by memcpy(..., len) writes past the SDS allocation before any parser can reject the body. Clamp this preallocation to at least len and check the + 5 overflow before copying.

Useful? React with 👍 / 👎.

@cosmo0920 cosmo0920 Jun 22, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

^ @yadoer

}
Comment thread
coderabbitai[bot] marked this conversation as resolved.
else {
initial_size = len;
}

if (initial_size < len) {
stream->status = HTTP_STREAM_STATUS_ERROR;
return -1;
}

if (http2_request_body_limit_exceeded(stream, initial_size)) {
stream->status = HTTP_STREAM_STATUS_ERROR;

return -1;
}

stream->request.body = cfl_sds_create_size(initial_size);

if (stream->request.body == NULL) {
stream->status = HTTP_STREAM_STATUS_ERROR;
Expand Down
Loading