This is part of fix to SWDEV-208805. The other part will
be in the build_* script.
Change-Id: I36397e3f918d08170db8bb228722a2b7389af83b
[ROCm/amdsmi commit: 0e5c44de2a]
* Update doc. on api-support function
* Check for valid integer value when reading a monitor int. val.
* If fan-write test attempts to set speed higher than max.
possible, then skip the test
Change-Id: I01ad0ab1f4caffdb0d2c26e9575f278c35a6b017
[ROCm/amdsmi commit: 52dfa4bcca]
For device-getter functions, allow users to specify a nullptr
for the provided buffer. In those cases, the function will return
RSMI_STATUS_NOT_SUPPORTED if the hardware or system software does
not support the function. If the function is supported, then
RSMI_STATUS_INVALID_ARGS will be returned, unless a different
error is encountered.
Additionally, tests and documentation were updated to reflect
this change.
Change-Id: Ie7db3a4c8c66af97ebd7ee1e3b95cd331ace9d9c
[ROCm/amdsmi commit: 68d25e82fd]
Add support and testing for reading the vram vendor associated with
the GPU. The vram vendor can be found as a separate sysfs file at:
/sys/class/drm/card[X]/device/mem_info_vram_vendor
The vram vendor is displayed as a string value.
Change-Id: I12c8e56e57f45aa08d7d6c25338c4e468ed1c7fc
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
[ROCm/amdsmi commit: 2412dff6a2]
The new functions added in this commit allow a caller to tell up
front what functions, function variants and monitors are
supported.
Also,
* fixed a few documentation/formatting issues
* fixed a process_info test issue
Change-Id: I2184ab1a4a6898f847e791f273e2185d556e78e9
[ROCm/amdsmi commit: 551b15182b]
If the 32-bit domain is found in the kfd node properties for
a device, then it will be used when constructing the bdfid.
If it's not present, it will continue to use the 16 bit version.
Also, whether or not 32b or 16b are used for the domain, the
domain will now be placed in the upper 32b of the 64b bdfid.
* Fixed some unrelated doxygen issues
Change-Id: Icb5116daa1ab45ee305bdbe6cd5df5736dd3ffa3
[ROCm/amdsmi commit: 469af303d6]
* Specifically, address case when brand name is longer than buffer
provided
* Also, slightly modify prototype to match similar, existing APIs.
* Address some cpplint issues.
Change-Id: Iaf77304e23085123e88f301e4b33bc4e6be2a225
[ROCm/amdsmi commit: 01e0800741]
Add support and testing for reading the brand name associated with
a specific GPU (such as mi25, mi50, mi60, etc). The brand name is
associated with the SKU of the GPU, and some brand names can be
mapped from multiple different SKUs.
Change-Id: I36eb95ca8e72efdd294ccd684841195925dfe820
Signed-off-by: Ori Messinger <Ori.Messinger@amd.com>
[ROCm/amdsmi commit: 7f2d970a80]
Also, use abbreviated ROCM_BUILD_ID environment variable for job
and build number, if it's available.
Change-Id: Ib5a721f5920f1008bb6382935f7b439429389de0
[ROCm/amdsmi commit: aa2db48237]
Library version will now only have major and minor. Package
version will now include number of commits since previous
package. Both SO and package versions rely on git tags to
determine the current build and the commits since the last
release.
Change-Id: If2bda74bf342930a9e07f5c91cb1380b6b7c64ca
[ROCm/amdsmi commit: fe738eaedb]
RAS formatting changed, so get it to handle both types of sysfs output
until it's normalized
Change-Id: I56f2a2495af8ff4d01011bc614283376afb9ad0a
[ROCm/amdsmi commit: a34832f11e]
Also, don't return an error for empty sysfs files. The reserved memory
page file will often have no lines. We don't want it to appear that
this function is not supported if the file is empty.
Change-Id: I1d28bb184ea587bb578fe71dd75adc2a812d09a8
[ROCm/amdsmi commit: 73c54e1fd0]
Function to get the drm minor number associated with ROCm device
Change-Id: I9356b9ca75151882acbb075076bc072f08b73aae
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
[ROCm/amdsmi commit: 68cb303a44]
Added implementation of and tests for
rsmi_dev_compute_process_info_by_pid_get() and
rsmi_dev_compute_process_info_get()
Change-Id: I4c4f5f39fe6701da37916c9ad41449b5d35ac7af
[ROCm/amdsmi commit: 9b93cbe21d]
Add support and testing for reading the Unique ID associated with a
specific GPU. This ID will persist across reboots, even if the GPU is
moved to a different machine. Note that this is per-GPU, not per-card,
as some cards have multiple GPUs, and each GPU will get a unique
identifier
Change-Id: Idce50c6febc2ceb1a4c1200d2489ec8b9d8fe174
[ROCm/amdsmi commit: 35d2807196]
* If vendor/device/subsystem name is not found, use device ID string
* Update documentation for get-name functions
* Add support for junction, edge and memory temperature sensors
[ROCm/amdsmi commit: 11f714326b]
With newly added initialization parameters that can be
passed to rsmi_init(), you can tell RSMI to consider other
devices.
Also:
-fixed incorrect header file name that would break in C
builds
-modified rsmi_init() and rsmi_shut_down() to reinitialize and
clear static structures
[ROCm/amdsmi commit: ae7ca83920]
With newly added initialization parameters that can be
passed to rsmi_init(), you can tell RSMI to consider other
devices.
Also:
-fixed incorrect header file name that would break in C
builds
-modified rsmi_init() and rsmi_shut_down() to reinitialize and
clear static structures
[ROCm/amdsmi commit: d7de7e23cb]
* By default, only consider AMD GPU's in RSMI device list
With newly added initialization parameters that can be
passed to rsmi_init(), you can tell RSMI to consider other
devices.
Also:
-fixed incorrect header file name that would break in C
builds
-modified rsmi_init() and rsmi_shut_down() to reinitialize and
clear static structures
* By default, only consider AMD GPU's in RSMI device list
With newly added initialization parameters that can be
passed to rsmi_init(), you can tell RSMI to consider other
devices.
Also:
-fixed incorrect header file name that would break in C
builds
-modified rsmi_init() and rsmi_shut_down() to reinitialize and
clear static structures
* Remove duplicate definition of rsmi_init_flags_t
[ROCm/amdsmi commit: a0014f34df]
With newly added initialization parameters that can be
passed to rsmi_init(), you can tell RSMI to consider other
devices.
Also:
-fixed incorrect header file name that would break in C
builds
-modified rsmi_init() and rsmi_shut_down() to reinitialize and
clear static structures
[ROCm/amdsmi commit: 15ca1c26a0]
With newly added initialization parameters that can be
passed to rsmi_init(), you can tell RSMI to consider other
devices.
Also:
-fixed incorrect header file name that would break in C
builds
-modified rsmi_init() and rsmi_shut_down() to reinitialize and
clear static structures
[ROCm/amdsmi commit: 2d847e3afd]
With newly added initialization parameters that can be
passed to rsmi_init(), you can tell RSMI to consider other
devices.
Also:
-fixed incorrect header file name that would break in C
builds
-modified rsmi_init() and rsmi_shut_down() to reinitialize and
clear static structures
[ROCm/amdsmi commit: 1dfef717bb]
Also, added code to destroy/recreate mutex if we can't get a lock
within 3 seconds, when shared memory mutex is initialized.
[ROCm/amdsmi commit: 34c977bd06]
This commit uses a pthread mutex in shared memory to prevent
almost all cases of multiple processes simultaneously
reading/writing to device sysfs files. The main existing race
condition is when 2 processes are starting at the same time,
setting up their shared memory and mutexes. Since this is meant
to prevent collisions among thread and processes, the small
shared memory segments (big enough for a pthread_mutex) will
persist until reboot.
[ROCm/amdsmi commit: 5e24a77193]
Chris Freehill
Ori Messinger
Kent Russell
Harish Kasiviswanathan
Chris Freehill