This won't work for obvious reasons, so exit with an error instead of
trying to access a file that doesn't exist and segfaulting
Change-Id: Id1230922fa6e9a19e9394280faad88a43c7d2e34
The rsmi_dev_temp_metric_get() can also support the HBM
temperatures which is retrieved from gpu_metrics.
Change-Id: I96b979296e90cf881523627b41b1a02849676416
Previously, RSMI assumed that the event counter values returned
from perf were only new events. But in fact, when we read the
counter values, they are running totals. To account for this, we
now record the value we read and take the difference between the
current value and the previously recorded value.
Change-Id: I1e04b514e89c7c4d4719889f2dae3a1283864e7f
rocm_smi.py --set<m|s>clk was treating the freq as a string.
This causes problems in parsing when the index is more than 1
digit. Now, treat the indexes as integers.
Change-Id: Ia0d859d33b685fe90689a86ff1c83980808b1514
Previously, during the rsmi_init discovery process, the existence
of an hwmon# directory was used to distinguish between gpus nodes
and non-gpu nodes. This isn't reliable in some scenarios. Instead,
the existence of the vbios_version file is used as an
indicator that the node is indeed a gpu.
Change-Id: Icfbe5c42ed0970077b05f25c3d209308a31bec85
The purpose of this patch is to fix a power cap bug for --setpoweroverdrive.
This bug occurs when the user attempts to set a lower wattage than the current
or default wattage, which displays an unnecessary warning message.
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: I730d2c6031b7d7c4af5acf32ecd28da5ca21ab12
The purpose of this patch is to implement GPU reset functionality
in the LIB, and to call it from the rocm_smi python CLI.
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: Iaf525f7016f8354a7fd93af0209ca2e97ef4fd56
The purpose of this patch is to fix a fan speed bug for --showfan.
This bug occurs when the current and/or maximum fan speeds are not
found by the LIB, which displayed an unclear error message.
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: Ied06e460f22391238dd2d86572813e2a5a64f45b
static-libasan doesn't exist, so use the easier-to-remember
shared-libsan and change static-libasan to static-libsan
Signed-off-by: Kent Russell <kent.russell@amd.com>
Change-Id: Ieef480aacdd770f3bb40673a2e8f8306b308b1c9
Instead of looking solely in ../lib, try looking in any /opt folder as a
backup option. This is a little more robust and hopefully leads to fewer
issues trying to find the lib
Change-Id: Ie0d3944b48b32d9965917e5c831388838b6d4ef7
If we fail to find an expected temperature or voltage label
file, previously we were attempting to re-add a mapping of file
index to sensor types. Attempting to insert a map item that is already
present has no effect, so there should be no functional change.
This was a remnant of old code that should have been deleted.
Change-Id: Ie6f8a62f619a1ae58756e0fd891532434518cf78
The environment variable RSMI_DEBUG_INFINITE_LOOP is introduced
to facilitate debugging RSMI in user applications. When this
env. variable is non-zero, an infinite loop will be entered in
rsmi_init(). At this point, a debugger can be attached and RSMI
can be debugger. This only applies to debug builds.
Change-Id: I23f6dd730fc965764295070de053314a1cc5b6aa
There are some systems that don't have sudo, and since we require sudo
for any of the "set" functionality, add it to "Suggests".
See https://github.com/RadeonOpenCompute/ROCm/issues/1245
Change-Id: I9428b9a68810ee8b51f91bb2e3b63312463161b0
Now that rocm-smi is deprecated, change the DEB/RPM info so that it
provides the rocm-smi package. This will allow for a seamless transition
over during ROCm upgrades
Change-Id: Ia29aab6e45c5974f7b623b786d0649710ba1f7cc
The purpose of this patch is to check if each valid clock is supported
on the GPU before attempting to retrieve its value.
The valid clocks are: dcefclk, fclk, mclk, pcie, sclk, socclk.
This should get rid of the 'one or more commands failed' message when
running --showclkfrq or --showclocks on a machine that doesn't support
all the possible valid clocks.
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Change-Id: I1fb10989fc1a36f38b68a23e17e6e600ed0ac85b
The purpose of this patch is to add Json functionality to showPids
by modifying the print2DArray function to use printSysLog.
Change-Id: Ie834d209b29332777c3f13f776f81c37d94b01b6
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Enums referenced in the test did not match what's in rocm_smi.h.
Added static assert to try to catch this. Also moved enum string
map to test_common.cc/h where other such maps are.
Also, fixed some cpplint issues.
Change-Id: I683553248ceb2fabb28ce1a1208bc9744aaf88d6
rocm_smi.py had an issue where it gets process information
in 2 different places. If the process disappears in between
those 2 places, a crash would occur.
This fix gracefully returns in this scenario.
Reading the file information from /proc instead of using
the python subProcess() call was considered, but it has the
drawback of not being able to read the process names of
processes not owned by the caller.
Change-Id: If812c4641f00da37e99defb0740f670107c8a797
Some rsmi apps fail without much explanation when
rsmi_init() fails. This patch hopes to provide some clues to
the reason for the failure.
Change-Id: Id51308dc327b9871d537dd3e709b677db4ef10bc
Previously, when a process holding a shared mutex was killed,
the next time an RSMI application was started, it would not be
able to obtain the mutex--the application would have to exit.
This fix uses pthread_mutexattr_setrobust() to detect this
situation and act accordingingly.
Also, add some missing, needed mutexes and move mutexes
closer to where the protect resource is used.
Change-Id: Icfdc3a246f4cfa3fd008e3f13472199abd76fd35
A special mode of operation to achieve minimal performance variation by letting
the user have the ability to provide the desired frequency to be set as the soft limit.
The user can control the entry and exit to the mode via rocm-smi a mechanism to
enter / exit performance determinism mode as below.
Enter performance determinism mode:
- hold a lock
- write performance_determinism to power_dpm_force_performance_level
- write input clk_freq to pp_dpm_sclk
- release lock
Exit performance determinism_mode:
- hold a lock
- write auto to power_dpm_force_performance_level
- release lock
Signed-off-by: Divya Shikre <DivyaUday.Shikre@amd.com>
Change-Id: Ia24e27954cdf1c4337ffc83d8948fbdfaf4552d2
The purpose of this patch is to fix a bug present when using the
--gpureset option on a machine that has both an AMD GPU and a
non-AMD GPU (such as a motherboard's integrated graphics).
This bug occurs due to non-AMD GPUs being ignored by the LIB when
enumerating a list of valid AMD GPUs, causing the gpuReset method
to attempt a reset on the integrated graphics.
Change-Id: I1c03a3c41f905786e3c8246ec0c2b42786ff1770
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Update the event notification tests to handle both GPU pre reset
and GPU post reset events. GPU post reset event takes sometime to
be generated after the pre reset event, so issue another
notification read to wait for post reset event.
Change-Id: I2812760b184d5357130e478cc35d27b14592abb3
The purpose of this patch is to fix a bug present when using the
--showproducname option, resulting in the following error:
undefined symbol: rsmi_dev_sku_get
This bug fix uses a substring from vbios version instead of using the
LIB's rsmi_dev_sku_get to avoid getting the undefined symbol error.
Change-Id: I56d72a481d5dde44c56106ae297f4bcff40ac15f
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
This implementation is copied directly from the previous rocm_smi.py
script; This feature is experimental and will be updated or removed with
feauture releases.
Signed-off-by: Elena Saknovitch
Change-Id: I5cd38266946302bc4123aeafaa825e13f704235e
The purpose of this patch is to add CU occupancy functionality to showPids
by calling rsmi_compute_process_info_get from the LIB.
Now showPids shows the following information on (KFD compute) processes:
PID, process name, GPU(s), VRAM used, SDMA used, and CU occupancy.
Change-Id: Ie005901e0eb946ef0fbb3523245ca451c4eed595
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
Bill(Shuzhou) Liu
Kent Russell
Divya Shikre
Cole Nelson
Chris Freehill
Ori Messinger
Elena
Ashutosh Mishra
Mukul Joshi
Ramesh Errabolu