The current codebase makes use of one-element arrays in the following
form:
struct something {
int length;
u8 data[1];
};
struct something *instance;
instance = kmalloc(sizeof(*instance) + size, GFP_KERNEL);
instance->length = size;
memcpy(instance->data, source, size);
but the preferred mechanism to declare variable-length types such as
these ones is a flexible array member[1][2], introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on. So, replace
the one-element array with a flexible-array member.
Also, make use of the new struct_size() helper to properly calculate the
size of struct SISLANDS_SMC_SWSTATE.
This issue was found with the help of Coccinelle and, audited and fixed
_manually_.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The current codebase makes use of one-element arrays in the following
form:
struct something {
int length;
u8 data[1];
};
struct something *instance;
instance = kmalloc(sizeof(*instance) + size, GFP_KERNEL);
instance->length = size;
memcpy(instance->data, source, size);
but the preferred mechanism to declare variable-length types such as
these ones is a flexible array member[1][2], introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on. So, replace
the one-element array with a flexible-array member.
Also, make use of the new struct_size() helper to properly calculate the
size of struct NISLANDS_SMC_SWSTATE.
This issue was found with the help of Coccinelle and, audited and fixed
_manually_.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Just check if it's an APU. The checks for the ppfuncs are
pointless because if we don't have them we can't power up
sdma anyway so we shouldn't even be in this code in the first
place. I'm not sure about the in_gpu_reset check. This
probably needs to be double checked. The fini logic doesn't
match the init logic however with that in_gpu_reset check
in place which seems odd.
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Just register the a pointer to the backlight device and use
that. Unifies the DC and non-DC handling.
Acked-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Always use the PCI GART instead. We just have to many cases
where AGP still causes problems. This means a performance
regression for some GPUs, but also a bug fix for some others.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
When smu_i2c_eeprom_init is called on the smu resuming process
under sroiv mode, there will be a call trace:
[ 436.377690] dump_stack+0x63/0x85
[ 436.377695] kobject_init+0x77/0x90
[ 436.377704] device_initialize+0x28/0x110
[ 436.377708] device_register+0x12/0x20
[ 436.377756] i2c_register_adapter+0xeb/0x400
[ 436.377763] i2c_add_adapter+0x5a/0x80
[ 436.377951] arcturus_i2c_eeprom_control_init+0x60/0x80 [amdgpu]
[ 436.378123] smu_resume+0xcc/0x110 [amdgpu]
[ 436.378247] amdgpu_device_gpu_recover+0xfb1/0xfc0 [amdgpu]
[ 436.378401] amdgpu_job_timedout+0xf2/0x150 [amdgpu]
[ 436.378414] drm_sched_job_timedout+0x70/0xc0 [amd_sched]
[ 436.378420] ? drm_sched_job_timedout+0x70/0xc0 [amd_sched]
[ 436.378430] process_one_work+0x1fd/0x3f0
[ 436.378438] worker_thread+0x34/0x410
[ 436.378444] kthread+0x121/0x140
[ 436.378451] ? process_one_work+0x3f0/0x3f0
[ 436.378456] ? kthread_create_worker_on_cpu+0x70/0x70
[ 436.378464] ret_from_fork+0x35/0x40
This is because smu_i2c_eeprom is not released on gpu recovering.
Actually, smu_i2c_eeprom_init/fini are only needed under bare
mental mode.
Signed-off-by: Hua Zhang <hua.zhang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
by default, vega20 will use legacy powerplay driver.
in order to maintain the code conveniently in the future,
remove the support of vega20 from swsmu.
Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
the vega20 asic uses legacy powerplay driver by default.
1. cleanup is_support_sw_smu_xgmi() function.
(only use for vega20 xgmi pstate check)
2. by default, the vega20 set xgmi pstate by legacy powerplay routine.
Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Change memory training init and finit a common function, as it only have
software behavior do not relay on the IP version of PSP.
Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This change removes internal rounding in dml_log2 function.
Dml_log2 is expected to return a float output. In case an int is needed
dml will floor the output on it's own.
Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Region 4 is non cacheable and slower than using cache window 4.
[How]
Check the firmware version to determine how we should program the
base address and memory windows.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
In order to switch over the inbox from region4 to cw4 we need to know if
the firmware is capable of properly invalidating the cache before
reading the commands.
Easiest way is to just check the firmware version, but we don't have the
helper macros or a way for the dmub_srv to know what version it is.
[How]
Add a new fw_version field to the creation parameters that driver can
optional pass in. Assumes a version of 0x00000000 is invalid.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[WHY]
Currently we're copying the entire bios image into vbios. Loading time
for FW with entire bios(54272 bytes) is 105138us. By copying only the
sections of bios we're using(4436 bytes), loading time drops to 104326us
which saves us 812us.
[HOW]
ROM header, master data table, and all data tables will be packed in
contiguous manner. The offsets for the data tables are remapped to their
newly packed location.
Signed-off-by: Jake Wang <haonan.wang2@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
DP link layer CTS specs updated to change the test parameters in test
4.2.1.1.
Before it requires source to delay 400us on aux no reply.
With the specs updates Errata5, it requires source to delay 3.2ms
(based on LTTPR aux timeout)
This causes our test to fail after updating with the latest test
equipment firmware.
[how]
the change is to allow LTTPR 3.2ms aux timeout delay by default.
And only set to 400us if LTTPR is not present.
Before this piece of logic is interwined with LTTPR support.
Now we will default to 3.2ms aux timeout even if LTTPR support is not
enabled by driver.
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Due to packing of abm_config_table, memory addresses aren't aligned to
32 bit boundary dmcub prefers. Therefore when using pointers to this
structure, it's possible that dmcub will automatically align the data
read from that address, yielding incorrect values.
[How]
Instead of packing 1 byte boundary, explicitly pack values to 4 byte
boundary. Since there is a dependency on the existing iram table
structure on driver side, we must copy to a second structure, which is
aligned correctly, before passing to fw.
Signed-off-by: Wyatt Wood <wyatt.wood@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Christian Koenig pointed out a code duplication related to bit swap in
case of big-endian manipulation. This commit adds a helper for handling
this verification and reduces the requirement of replicate some part of
the code.
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Reviewed-by: Wyatt Wood <Wyatt.Wood@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
If bss_data_size is 0 then we shouldn't be passing down fw_bss_data into
the DMUB service since the region isn't really "valid."
[How]
Pass NULL instead if the size is 0.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Zhan Liu <Zhan.Liu@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
New unified firmware binary with only inst const still passes down
fw_bss_data != NULL and params->bss_data_size == 0 from DM.
This leads it into the legacy path causing firmware state allocation to
be too small.
[How]
Check bss_data_size as well.
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Zhan Liu <Zhan.Liu@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[WHY]
Failing validation when building scaling parameters causes corruption to
occur due to pipe splitting with smaller pixel widths than HW supports.
This needs to fail silently for now to hide the corruption until the
corruption itself can be fixed.
[HOW]
Do not fail validation if building scaling params fails.
Signed-off-by: Sung Lee <sung.lee@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Remove dm_write_persistent_data and dm_read_persistent_data as
persistence should be handled in DM.
[How]
Remove functions. Move read/write calls into DM layer while maintaining
logic.
Signed-off-by: Jaehyun Chung <jaehyun.chung@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We should be able to skip restoring LOCAL (user) binds within the GGTT on
resume and let them be restored upon demand. However, our consistency
checks demand that the bind flags match the node state, and we cannot
simply clear the flags, we need to evict as well. For now, make sure we
restore the bind flags exactly upon resume.
Fixes: 0109a16ef3 ("drm/i915/gt: Clear LOCAL_BIND from shared GGTT on resume")
Fixes: bf0840cdb3 ("drm/i915/gt: Stop cross-polluting PIN_GLOBAL with PIN_USER with no-ppgtt")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200528150452.7880-1-chris@chris-wilson.co.uk
We have a I915_REQUEST_NOPREEMPT flag that we set when we must prevent
the HW from preempting during the course of this request. We need to
honour this flag and protect the HW even if we have a heartbeat request,
or other maximum priority barrier, pending. As such, restrict the
timeslicing check to avoid preempting into the topmost priority band,
leaving the unpreemptable requests in blissful peace running
uninterrupted on the HW.
v2: Set the I915_PRIORITY_BARRIER to be less than
I915_PRIORITY_UNPREEMPTABLE so that we never submit a request
(heartbeat or barrier) that can legitimately preempt the current
non-premptable request.
Fixes: 2a98f4e65b ("drm/i915: add infrastructure to hold off preemption on a request")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200527162418.24755-1-chris@chris-wilson.co.uk
[Why]
If VUPDATE_END is before VUPDATE_START the delay calculated can become
very large, causing a soft hang.
[How]
Take the absolute value of the difference between START and END.
Signed-off-by: Aric Cyr <aric.cyr@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Pull cgroup fixes from Tejun Heo:
- Reverted stricter synchronization for cgroup recursive stats which
was prepping it for event counter usage which never got merged. The
change was causing performation regressions in some cases.
- Restore bpf-based device-cgroup operation even when cgroup1 device
cgroup is disabled.
- An out-param init fix.
* 'for-5.7-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
device_cgroup: Cleanup cgroup eBPF device filter code
xattr: fix uninitialized out-param
Revert "cgroup: Add memory barriers to plug cgroup_rstat_updated() race window"
gcc-9 gets confused by the code flow in check_dirty_whitelist:
drivers/gpu/drm/i915/gt/selftest_workarounds.c: In function 'check_dirty_whitelist':
drivers/gpu/drm/i915/gt/selftest_workarounds.c:492:17: error: 'rsvd' may be used uninitialized in this function [-Werror=maybe-uninitialized]
I could not figure out a good way to do this in a way that gcc
understands better, so initialize the variable to zero, as last
resort.
Fixes: aee20aaed8 ("drm/i915: Implement read-only support in whitelist selftest")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20200527140526.1458215-2-arnd@arndb.de
Conditional spinlocks make it hard for gcc and for lockdep to
follow the code flow. This one causes a warning with at least
gcc-9 and higher:
In file included from include/linux/irq.h:14,
from drivers/gpu/drm/i915/i915_pmu.c:7:
drivers/gpu/drm/i915/i915_pmu.c: In function 'i915_sample':
include/linux/spinlock.h:289:3: error: 'flags' may be used uninitialized in this function [-Werror=maybe-uninitialized]
289 | _raw_spin_unlock_irqrestore(lock, flags); \
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/i915/i915_pmu.c:288:17: note: 'flags' was declared here
288 | unsigned long flags;
| ^~~~~
Split out the part between the locks into a separate function
for readability and to let the compiler figure out what the
logic actually is.
Fixes: d79e1bd676 ("drm/i915/pmu: Only use exclusive mmio access for gen7")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20200527140526.1458215-1-arnd@arndb.de
