Nevermind, this change is already in https://github.com/apache/httpd/pull/366, so if we bring that in, all should be well.
> Am 22.08.2023 um 13:48 schrieb Stefan Eissing via dev <dev@httpd.apache.org>: > > I believe we can add that to the backport proposal without resetting the > votes. > >> Am 22.08.2023 um 12:57 schrieb Ruediger Pluem <rpl...@apache.org>: >> >> Currently 2.4.x fails to compile for me with the following error: >> >> In file included from h2_c1_io.c:24: >> h2_c1_io.c: In function ‘pass_output’: >> /usr/src/apache/httpd-2.4.x/include/http_log.h:495:14: warning: ‘rv’ may be >> used uninitialized in this function >> [-Wmaybe-uninitialized] >> ap_log_cerror_(file, line, mi, level, status, c, __VA_ARGS__); \ >> ^~~~~~~~~~~~~~ >> h2_c1_io.c:264:18: note: ‘rv’ was declared here >> apr_status_t rv; >> ^~ >> >> >> The following patch fixes this. It is part of r1910507 which is not >> backported yet. >> >> Index: modules/http2/h2_c1_io.c >> =================================================================== >> --- modules/http2/h2_c1_io.c (revision 1911839) >> +++ modules/http2/h2_c1_io.c (working copy) >> @@ -267,7 +267,7 @@ >> /* recursive call, may be triggered by an H2EOS bucket >> * being destroyed and triggering sending more data? */ >> AP_DEBUG_ASSERT(0); >> - ap_log_cerror(APLOG_MARK, APLOG_ERR, rv, c, APLOGNO(10456) >> + ap_log_cerror(APLOG_MARK, APLOG_ERR, 0, c, APLOGNO(10456) >> "h2_c1_io(%ld): recursive call of h2_c1_io_pass. " >> "Denied to prevent output corruption. This " >> "points to a bug in the HTTP/2 implementation.", >> >> >> Not sure how we proceed best from here. If we commit it to 2.4.x (after >> appropriate voting of course) it would break the current >> backport proposal. But as the backport proposal is a PR it might be easy to >> take out this small part of the proposal. >> >> Regards >> >> Rüdiger >
