These registers are shared between BR and BV. This means they can only
be written in BV or in BR during a special "disable binning" section
which synchronizes BR and BV. The register stomping infrastructure isn't
aware of this, so disable stomping for VSC.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37722>
When the system is under memory pressure (which can happen, for
example, during CI runs), don't immediately give up the exec ioctl
(which, for Vulkan, will result in the device being declared lost).
Instead, retry a little bit just like we do for i915.ko.
This is a trade-off.
One of the reasons to *not* have unified behavior regarding ENOMEM
between i915.ko and xe.ko is the fact that xe.ko uses vm_bind, so if
the user tried to bind more memory than it is able to, we'll just keep
getting ENOMEM as long as we retry the ioctl. We now have a retry
limit, so we'll eventually return the error.
On the other hand, if the problem is other applications consuming all
the memory, having the retry loop may really help avoid unnecessarily
marking the device as lost, since one of our retries may eventually
succeed.
I believe the tradeoff of "we'll now eventually succeed in some cases
where it's possible to succeed, at the expense of retrying for a few
seconds until giving up in cases where we would never be able to
succeed" is an improvement.
If xe.ko ever gives us a way to differentiate between the two
different reasons for ENOMEM, we'll be able to make things much
better. We can also tune our timeouts if needed.
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37559>
If nothing has freed memory until that point, return the error, which
may make the upper layers report the device as lost. It could be that
the system is under very very heavy swapping and that waiting a little
more would make it work, but let's try 16s for now.
v2: Bring down the timeout from ~60s to ~16s (José).
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37559>
If the ioctl is returning ENOMEM, incessantly retrying does not seem
to be the best way to proceed. After the second retry, sleep 0.1ms,
then more each time, giving the CPU some time to run the other threads
and processes, in the hope that whatever is eating all the memory
might eventually return it.
If the problem is the current thread, then busy looping won't help
either, so here we at least save some power before the user kills the
app.
v2: Adjust the control flow and the sleep time.
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37559>
We were walking the instructions in the block for each
first-rpt-instruction in the block. Instead, on the first query per
block, make a set of all the rpts in the block, so we can O(1) check for
the remainder.
shader-db runtime for deadspace3 -7.60909% +/- 2.28996% (n=10) on a
debugoptimized build.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37625>
Android has to enable dumping globally. There's no per app based env
var at runtime since most apps just fork from Zygote process. So we have
to add process name to the dump file name. Now with pandecode.dump as
the base name, it'll be like below on Android:
- pandecode.dump.com.example.VkCube.ctx-*
- pandecode.dump.com.google.android.apps.nexuslauncher.ctx-*
This can be generally useful on Linux as well when debugging different
things to avoid accidentally touching existing dumps.
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37729>
A 64-bit atomic load/store should be considered entirely out-of-bounds if
any part of it is out-of-bounds. Since we implemented these as 32-bit vec2
load/store, it would have been possible for the first half to be in-bounds
while the second half is out-of-bounds.
From 9.6.1. Robust Buffer Access of Vulkan 1.4.324 specification:
> Any non-atomic access to a uniform, storage, uniform texel, or storage
> texel buffer wider than 32-bits may be treated as multiple 32-bit
> accesses that are separately bounds checked.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36602>
Due to the division of TU_DEBUG options into runtime and envvar
options, it limited where options could be set from when
TU_DEBUG_FILE was being used. This commit addresses that by allowing
the envvar to set runtime debug options even when TU_DEBUG_FILE is
active while also allowing the file to set non-runtime options if
the file included them at startup.
Signed-off-by: Dhruv Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37580>
The primary CS doesn't need to use chaining in order to use IB2.
Allow using IB2 packets when chaining is disabled.
Rationale for this patch:
When chaining is enabled (the default), this patch removes a
useless check.
When chaining is disabled (by noibchaining), this patch allows us
to use IB2 without chaining.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37280>
All CS always use IBs, so the naming was confusing.
Rename these fields to chain_ib to better reflect
what it actually means, which is enabling chaining:
radv_amdgpu_winsys::use_ib_bos
radv_amdgpu_cs::chain_ib
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37280>
Vinson Lee