| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
Signed-off-by: Jakov Smolić <jsmolic@gentoo.org>
|
|
|
|
| |
Signed-off-by: Jakov Smolić <jsmolic@gentoo.org>
|
|
|
|
|
|
|
|
|
| |
Still having a hard time to get this to work on my hardware,
but Fedora is pushing that option and trying to enable it by
default now. May be good to give an explanation so users can
more easily choose if they can ignore the warning or not.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
| |
Keep 495.44-r2 and 495.46-r10 for now in case of regressions, but
-r0 and -r20 should have no real uses.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Removed dbus workarounds, should hopefully be fixed now which should
also fix bug #828433
Bump NV_KERNEL_MAX to 5.16 as it fixes the known build issue with
CONFIG_DEVICE_PRIVATE.
New dependency on libpciaccess for nvidia-xconfig, but not going to
make this a separate USE unlike persistenced (or at least for now)
given Xorg depends on this library as well and fits well with USE=X.
Newly adds nvidia-powerd.service, but it's not documented, haven't
got it to work on my hardware and source appears unavailable at the
moment. Currently unsure what it's good for, but perhaps used to
manipulate power states based on OpenGL/Vulkan usage (which are known
to communicate with it).
Closes: https://bugs.gentoo.org/828433
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously choose to switch to <= from < to avoid confusion when
people see the version (i.e. users would see <5.16 and think up to
5.16 is supported), but feels kind of innacurate if taken literally
and so add .x at the end.
wrt 5.16.x, not bumping NV_KERNEL_MAX yet. It will work for most
people with custom kernels (except 0/460 and 0/vulkan), but will
fail if kernel has CONFIG_DEVICE_PRIVATE set (gentoo-kernel-bin
has this set) because of MIGRATE_PFN_LOCKED definition removal.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: David Seifert <soap@gentoo.org>
|
|
|
|
| |
Signed-off-by: David Seifert <soap@gentoo.org>
|
|
|
|
|
|
| |
Manually tested and working.
Signed-off-by: James Le Cuirot <chewi@gentoo.org>
|
|
|
|
| |
Signed-off-by: James Le Cuirot <chewi@gentoo.org>
|
|
|
|
| |
Signed-off-by: James Le Cuirot <chewi@gentoo.org>
|
|
|
|
|
|
| |
Manually tested and working.
Signed-off-by: James Le Cuirot <chewi@gentoo.org>
|
|
|
|
| |
Signed-off-by: James Le Cuirot <chewi@gentoo.org>
|
|
|
|
|
|
| |
Manually tested and working.
Signed-off-by: James Le Cuirot <chewi@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Removing old type ebuilds now that the newer ones are stable.
Still 495.44-r2 in ~arch but that one is kept for now due to a
495.46 regression unrelated to the ebuild. NVIDIA mentioned they're
looking into the issue so should hopefully be able to cleanup next
version (albeit recent development show it's possible to fix by
patching Xorg's xpresent -- whether it's a bug in Xorg that happened
to work by accident before or if it's a driver bug currently unclear).
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
| |
Should be no reason to wait longer for this, the ebuild changes
been tested and stabled through other branches already.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
While this applies to every ebuilds that build modules, nvidia-drivers
is often a user's first or only experience with these and support
requests from those that did not know are somewhat common (e.g.
upgraded kernel without rebuilding, or built modules against the
wrong kernel).
While wiki is better for a lot of these things, having more sources
mention this material increase chances to reach users.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
460.91.03 is unfortunately known needed for (at least) a few laptop
users, issues varying between not able to control backlight or not
seeing anything displayed at all (bug #780126 and bug #809482, along
with several threads on NVIDIA forums complaining about this).
Can't be kept eternally given NVIDIA shows no sign of continued 460
support and it requires xorg-1.20 plus <=5.10 LTS kernels (unlike even
the 390.xx branch) -- but, meanwhile, update ebuild to be like the
others and easier to maintain. Fortunately no known security issues.
Let's hope 470/495+ will receive fixes before removal is necessary
(perhaps to cleanup old xorg or security concerns), feedback from
users welcome to know when it works as expected or if anything else
can do (own hardware is not affected, so can't test).
Bug: https://bugs.gentoo.org/780126
Bug: https://bugs.gentoo.org/809482
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is not solved and still happening with 495.46 (doesn't seem
like anything can reasonably do about it from a packaging end,
and NVIDIA hasn't done anything), but the postinst message itself
wasn't shown unless updating 6+ months old nvidia-drivers with
coolbits in xorg.conf and is now mostly unused cruft.
If this is going to be a permanent state of affair, may be better
documented elsewhere and bug closed as an upstream issue (albeit
keeping open a bit longer for tracking).
Bug: https://bugs.gentoo.org/784248
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
| |
Done early'ish given 5.15.x kernel branch is stable now
and this version is needed to support it.
Closes: https://bugs.gentoo.org/830008
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Replaces the prebuilt libnvidia-egl-gbm with gui-libs/egl-gbm.
Unkeyworded mostly given the 495.xx situation is a mess right now
that may require users to pick specific revisions, and there's no
reason to bother users with one more unless they want to test it.
Should be officially rolled out next nvidia-drivers version bump.
Like egl-wayland, this is technically a optfeature but want
USE=wayland to give everything needed for decent wayland support.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
| |
Not quite clear if this is really an issue with xorg-server, but it is what it is.
Closes: https://bugs.gentoo.org/829628
Thanks-to: Remus-Gabriel Chelu <gaby_r_chelu@yahoo.es>
Signed-off-by: Sam James <sam@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
| |
Also update ebuild to match 495.46 changes.
Was tested with a 560ti + gentoo-kernel-bin-{5.10.86,5.15.10}
+ xorg-{1.20.14,21.1.2-r2} + plasma-5.23.4 without issues, this
driver branch still has some life in it.
Closes: https://bugs.gentoo.org/829635
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
| |
Closes: https://bugs.gentoo.org/829627
Signed-off-by: Sam James <sam@gentoo.org>
|
|
|
|
|
|
|
| |
495.46 is supposed to be a very trivial release, but there
may possibly be a regression. Restore for testing.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
| |
Closes: https://bugs.gentoo.org/827652
Signed-off-by: Sam James <sam@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
| |
This was primarily to raise awareness due to being masked (now isn't
except on plasma profile), other hints were just a small bonus which
don't really belong in the ebuild.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
>1.1.7 been masked up to now so it didn't matter much, but gbm-unaware
470 does not play well with gbm-aware egl-wayland-1.1.9 (notably with
Xwayland).
Hopefully won't be permanent given 470.xx will be around for several
years but, if 1.1.7 becomes a problem, 470's USE=wayland could be
dropped given EGLStream is likely to fall out of use.
Keeping the >-r1 restriction isn't necessary given -r0 had its own
blocker on nvidia-drivers[wayland].
For a slight annoyance, egl-wayland used to be a optfeature so
people may have it in their world file which will result in the
not-an-error-but-feels-like-one portage "WARNING: One or more..."
If reading this, users are invited to remove it from their world file.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
495.46 release is almost identical to 495.44 beside adding support
for a few additional cards, so the same issues (e.g. dbus spam) exist
which is patched in -r10 and left alone in -r0.
ebuild-wise, now reads .manifest to mimic nvidia-installer.
For a few advantages:
* less unused .so links or using scanelf to find sonames, matches
nvidia's intended links
* can bump without constantly adjusting for every new/removed files
* warns if something entirely new needs work (e.g. GBM support)
* more solid when using different .run, e.g. amd64 vs arm64, not
keyworded yet (waiting for confirmation that it works), but arm64
install broke in the past because of this
Other changes:
* IUSE=persistenced: add to control installing nvidia-persistenced
which allows libtirpc and acct-user/nvpd to be optional
(off by default given primarily intended for headless)
* nvidia-drm_gbm.so: install unconditionally to avoid spurious
errors with vaapi (no extra dependencies).
* nvidia-settings: hide gtk deprecation noise as it makes the log
hard to follow, considerably bigger, and the issue is already well
known (NVIDIA is hopefully due to revamp this tool, e.g. wayland).
* README.gentoo: add abi_x86_32 warning for new users given no longer
default, the "flip" elog may be removed in the future after users
had time to adapt with the profile change.
* drop multilib-build eclass as it doesn't make that much sense with
a binary-only package, compat32 libraries are still installed with
USE=abi_x86_32 (abi_x86_64 kept for compatibility with ebuilds
using ${MULTILIB_USEDEP} but is unused)
* fix build if /usr/bin/ld or LD points to ld.lld
* replaced .desktop and gtk2 seds by patches
Various other small adjustments/cleanups, changes will be rolled
out to older branches in time (patches are already based on 390.xx).
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
| |
Bit late but it's something users should know about (will at
least be shown when rebuilding for a new kernel).
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: Matt Turner <mattst88@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
| |
Closes: https://bugs.gentoo.org/827878
Signed-off-by: Sam James <sam@gentoo.org>
|
|
|
|
|
| |
Closes: https://bugs.gentoo.org/827872
Signed-off-by: Piotr Karbowski <slashbeast@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Profile had abi_x86_32 enabled by default for ~7 years (now removed),
and users may not necessarily have realized whether they needed it.
steam-overlay's steam-client-meta depends on nvidia's [abi_x86_32]
but wine* and other 32bit applications generally do not (typically
only check for virtual/opengl which only checks if mesa has it set,
or may check neither if depends on libglvnd directly).
Warn to miminize poor surprises to some extend.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Error message updated to help users enable it on kernel 5.15.3+
which no longer selects KMS_HELPER with CONFIG_DRM_FBDEV_EMULATION.
This may unfortunately cause a poor surprise (failing Xorg) to users
that did not notice this newly appearing message with minimal kernels
and didn't have a custom xorg.conf forcing nvidia.
Also modified sed to ensure nvidia-drm is properly built with support
if KMS_HELPER is a module and DRM is not (revbump latest arch/~arch
for this).
Bug: https://bugs.gentoo.org/824594
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
xorg-server-21+ isn't in tree yet, but done preemptively given
there will ABI version mismatch for xorg module preventing loading.
Probably simpler to check this from the nvidia ebuild than xorg.
390 - Still officially supported by NVIDIA for another year, but
no telling if/when NVIDIA will push an update for new Xorg. While
plans are to keep 0/390 for as long as possible, not planning to
let this block old xorg removal forever if NVIDIA does nothing
within the next year.
460 - This branch has little worth and is due for removal sooner
than later, only kept due to some leftover regressions in >=470.xx
affecting a few users.
--ignoreABI has not yet been tested, but would rather this route
be a last resort even if it works.
>=470 has no (known) problems with current xorg-server-9999.
PDEPEND used just in case given xorg meta and blockers involving
on nvidia-drivers could potentially get circular/messy.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|
|
|
|
| |
Signed-off-by: Matt Turner <mattst88@gentoo.org>
|
|
|
|
| |
Signed-off-by: Matt Turner <mattst88@gentoo.org>
|
|
|
|
| |
Signed-off-by: Georgy Yakovlev <gyakovlev@gentoo.org>
|
|
|
|
| |
Signed-off-by: Georgy Yakovlev <gyakovlev@gentoo.org>
|
|
|
|
| |
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
|