drm/amdgpu: fix IP instance memory leak on kobject add failure
From Guangshuo Li
7ca44cb14d0249992a8f1425ba004d6e2d16d255 in linux-6.18.y/6.18.56
f5a5add3605935163dd0f75749b6a7869e9765af in mainline linux
drm/radeon: Read the VRAM VBIOS signature with readb()
From Imre Kaloz
4d12b763c2c554571b5e03a04832c34b4e90e1d3 in linux-6.18.y/6.18.56
de1eab13efef83fd2b76f4688de5bf672644596f in mainline linux
drm/amdgpu/gfx6: fix firmware leak on teardown
From Willian Oliveira
652f631e2e3a8d3b11ad72dd0b8e59b5183de0d9 in linux-6.18.y/6.18.56
cc172e177795c5f87bc2d71768d077e1c10633e8 in mainline linux
drm/amd/pm/si: Fix updating clock limits on AC/DC
From Timur Kristof
36fa4159d005498e9527c617fd516ad3e81cee95 in linux-6.18.y/6.18.56
08c9100f65299c64439e22e6057d750ae0e6c660 in mainline linux
drm/amd/pm/si: Check AMD_IS_MOBILITY flag for PPSMC_SYSTEMFLAG_GPIO_DC
From Timur Kristof
e67aef159f430946bbfb6577c4aaaf130e3b9246 in linux-6.18.y/6.18.56
cd883d0a3937df82a5ea03709ba8ba623b868419 in mainline linux
drm/amd/display: check dc_state_create_copy() for NULL in dm_suspend
From Jiangshan Yi
197946faa1b7968046ba475ae2f266847c6f51ed in linux-6.18.y/6.18.56
015e72cde567d2450e282c2790773463156b2d0a in mainline linux
If eqn(7) or tbl(7) input is encountered in -mdoc -T markdown mode,
print a line describung the problem to the formatted output and also
raise an "unsupp" message, similar to what -T man already does,
instead of calling assert(3), which is never appropriate when dealing
with unsopported or incorrect input.
A conversation with kristaps@ reminded me of this issue.
Check for unsigned integer overflow when calculating how much beyond
our local window a peer has sent. Prevents them sneaking past the grace
we allow.
bz#4031, from rhowe425 at gmail.com, ok markus@
Remove the SSH_OLD_SESSIONID bug compatibility bit. It was only
set for SecureCRT 3.0 and SecureFX 1.7. Both date back to some
time prior to 2002 (vandyke's release history only goes back to
2002, and these releases predate that).
Prompted by rhowe425 at gmail.com via bz#4030, ok markus@
remove angr and angrop for now; this angr versiom isn't working with
python 3.14 and I haven't been able to figure out how to build new
versions. (a new major version has recently come out and might be worth
another attempt maybe when it's settled down a bit, but it looks like
some of the deps currently in devel/angr/xxx are no longer used so that
is probably worth redoing the set).
prompted by build failure after updating py-cparser reported by naddy,
but I noticed that it had already been failing at runtime for a while.
tb looked at trying to fix too.
ok jasper (maintainer)
tweak previous: drop stray blank line, drop a few words that accidentally
stayed behind, and fix some plural->singular issues;
feedback and OK mglocker@, and OK jca@ on an earlier version
Create the weak global _permit_setugid which must match the setuid/setgid
mode of the executable file. This is documented in execve(2).
The libc startup validation code to do this is not enabled yet.
xlock is setgid (auth), but does not use the bsd.prog.mk build infrastructure,
so it needs manual definition of "mode_t _permit_setugid = S_ISGID". The other
setgid program in X11R6 is xterm (group utmp), but it is built using bsd.prog.mk
and automatically picks up the _permit_setugid variable.
Update libexpat to version 2.8.5.
Changes relevant for OpenBSD have been backported earlier. Compiler
warning #1362 has been disabled, it does not work with gcc 4.2.1.
Move the hash table declarations into a separate header file as
this is needed for #1355. Convert salt test to non deprecated
XML_SetHashSalt16Bytes() function. Library bump is not necessary.
OK deraadt@
update to py3-dulwich-1.2.17
make a start at trying to allow rust/no-rust builds, but I'm not sure if
there's a good way to add SUPDISTFILES for the crates..
