https://bugs.kde.org/show_bug.cgi?id=523405

--- Comment #1 from Jan Iversen <[email protected]> ---
This is very likely a duplicate of bug 523105, which was reported a
few days earlier and identifies the root cause: the heif_reader
struct in HEIFHandler::ensureDecoder() is a block-local, and libheif
keeps only a raw pointer to it (StreamReader_CApi::m_func_table), so
the function table is read from a dead stack frame later on. That
matches my backtrace exactly - my crash is a jump to 0x0 from
StreamReader_CApi::request_range.

One difference worth noting: for me only a single HEIC file triggers
the crash, while bug 523105 reports it for every HEIC. That seems
consistent with the dangling-pointer analysis, where whether the
clobbered stack slot happens to hold garbage or a usable value is
essentially chance.

My report also confirms the regression boundary from a different
angle: Debian Trixie with kimageformats 6.27.0 is unaffected, two
openSUSE Tumbleweed machines with 6.28.0 both crash.

Feel free to close this as a duplicate.

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to