Add support to force specific module load
cover Add support to force specific module load
This patchset adds support for forcing a specific module to be loaded via command line. This feature can bring benefits for developers that need to create a new module since they can use IGT as a base test to guide their development (Test-driven development - TDD ). The first patch, expand the size of the module name accepted by IGT to make possible that modules with larger names can be loaded. The second patch has the force option implementation.
The discussion about this feature started with an RFC named "[PATCH i-g-t] [RFC] Add support to force specific module load", link:
https://www.spinics.net/lists/intel-gfx/msg166764.html
Rodrigo Siqueira (2):
Increase the string size for a module name
Add support for forcing specific module
lib/drmtest.c | 39 +++++++++++++++++++++++++++++++-------- lib/igt_core.c | 23 +++++++++++++++++++++++ lib/igt_core.h | 4 ++++ scripts/run-tests.sh | 4 +++- 4 files changed, 61 insertions(+), 9 deletions(-)
Signature · 6 lines
2.18.0 _______________________________________________ igt-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/igt-dev
PATCH 1/2 Increase the string size for a module name
Some modules name are larger than 5 characters, this can be a problem to add support for other modules. This patch, increase the maximum size in order to enable other modules to use IGT.
--- lib/drmtest.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
lib/drmtest.c +4 -2
@@ -60,6 +60,8 @@#include "ioctl_wrappers.h" #include "igt_dummyload.h" +#define NAME_LEN 32 +/** * SECTION:drmtest * @short_description: Base library for drm tests and tools
@@ -82,7 +84,7 @@ static int __get_drm_device_name(int fd, char *name)drm_version_t version;memset (&version,0 ,sizeof (version)); - version.name_len =4 ; + version.name_len =NAME_LEN ; version.name = name;if (!drmIoctl (fd,DRM_IOCTL_VERSION , &version)){
@@ -94,7 +96,7 @@ static int __get_drm_device_name(int fd, char *name)static bool __is_device (int fd,const char *expect) { -char name[5 ] ="" ; +char name[NAME_LEN ] ="" ;if (__get_drm_device_name (fd, name))return false;
Signature · 6 lines
2.18.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [PATCH i-g-t 1/2] Increase the string size for a module name
Quoted 32 lines · Rodrigo Siqueira
Some modules name are larger than 5 characters, this can be a problem to add support for other modules. This patch, increase the maximum size in order to enable other modules to use IGT.
Signed-off-by Rodrigo Siqueira <[email protected]>--- lib/drmtest.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)lib/drmtest.c +3
-1@@ -60,6 +60,8 @@#include "ioctl_wrappers.h" #include "igt_dummyload.h" +#define NAME_LEN 32 +/** * SECTION:drmtest * @short_description: Base library for drm tests and tools @@ -82,7 +84,7 @@ static int __get_drm_device_name(int fd, char *name)drm_version_t version;memset (&version,0 ,sizeof (version)); - version.name_len =4 ; + version.name_len =NAME_LEN ; version.name = name;if (!drmIoctl (fd,DRM_IOCTL_VERSION , &version)){
lib/drmtest.c static int __get_drm_device_name(int fd, char *name) · line 99
Quoted 5 lines
lib/drmtest.c +1
-1@@ -94,7 +96,7 @@ static int __get_drm_device_name(int fd, char *name)static bool __is_device (int fd,const char *expect) { -char name[5 ] ="" ; +char name[NAME_LEN ] ="" ;
This would need to be NAME_LEN + 1 to have room for the null-termination.
But, is this patch really necessary? Are there false matches from using a substring of 4 to match driver names?
Signature · 5 lines
Petri Latvala _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [igt-dev] [PATCH i-g-t 1/2] Increase the string size for a module name
Quoted 42 lines · Petri Latvala
Quoted 32 lines · Rodrigo Siqueira
Some modules name are larger than 5 characters, this can be a problem to add support for other modules. This patch, increase the maximum size in order to enable other modules to use IGT.
Signed-off-by Rodrigo Siqueira <[email protected]>--- lib/drmtest.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)lib/drmtest.c +3
-1@@ -60,6 +60,8 @@#include "ioctl_wrappers.h" #include "igt_dummyload.h" +#define NAME_LEN 32 +/** * SECTION:drmtest * @short_description: Base library for drm tests and tools @@ -82,7 +84,7 @@ static int __get_drm_device_name(int fd, char *name)drm_version_t version;memset (&version,0 ,sizeof (version)); - version.name_len =4 ; + version.name_len =NAME_LEN ; version.name = name;if (!drmIoctl (fd,DRM_IOCTL_VERSION , &version)){lib/drmtest.c static int __get_drm_device_name(int fd, char *name) · line 99
Quoted 5 lines
lib/drmtest.c +1
-1@@ -94,7 +96,7 @@ static int __get_drm_device_name(int fd, char *name)static bool __is_device (int fd,const char *expect) { -char name[5 ] ="" ; +char name[NAME_LEN ] ="" ;This would need to be
NAME_LEN+ 1 to have room for the null-termination.
Quoted 2 lines
But, is this patch really necessary? Are there false matches from using a substring of 4 to match driver names?
Hi,
To test the force option, I used two drivers: Bochs and VKMS. I noticed that Bochs failed to load because the module name is "bochs-drm". Additionally, if we want to add a force option, I assume that someone can have a module name larger than four characters. Make sense? Or Did I missed something?
Thanks for your feedback
Quoted 3 lines
Signature · 2 lines
-- Petri Latvala
Signature · 9 lines
Rodrigo Siqueira http://siqueira.tech Graduate Student Department of Computer Science University of São Paulo _______________________________________________ igt-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/igt-dev
Re: [Intel-gfx] [igt-dev] [PATCH i-g-t 1/2] Increase the string size for a module name
Quoted 43 lines · Rodrigo Siqueira
Quoted 42 lines · Petri Latvala
Quoted 32 lines · Rodrigo Siqueira
Some modules name are larger than 5 characters, this can be a problem to add support for other modules. This patch, increase the maximum size in order to enable other modules to use IGT.
Signed-off-by Rodrigo Siqueira <[email protected]>--- lib/drmtest.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)lib/drmtest.c +3
-1@@ -60,6 +60,8 @@#include "ioctl_wrappers.h" #include "igt_dummyload.h" +#define NAME_LEN 32 +/** * SECTION:drmtest * @short_description: Base library for drm tests and tools @@ -82,7 +84,7 @@ static int __get_drm_device_name(int fd, char *name)drm_version_t version;memset (&version,0 ,sizeof (version)); - version.name_len =4 ; + version.name_len =NAME_LEN ; version.name = name;if (!drmIoctl (fd,DRM_IOCTL_VERSION , &version)){lib/drmtest.c static int __get_drm_device_name(int fd, char *name) · line 99
Quoted 5 lines
lib/drmtest.c +1
-1@@ -94,7 +96,7 @@ static int __get_drm_device_name(int fd, char *name)static bool __is_device (int fd,const char *expect) { -char name[5 ] ="" ; +char name[NAME_LEN ] ="" ;This would need to be
NAME_LEN+ 1 to have room for the null-termination.
Quoted 7 lines
Quoted 2 lines
But, is this patch really necessary? Are there false matches from using a substring of 4 to match driver names?
Hi,
To test the force option, I used two drivers: Bochs and VKMS. I noticed that Bochs failed to load because the module name is "bochs-drm".
Driver name is still not the same as the kernel module name necessarily.
Quoted 3 lines
Additionally, if we want to add a force option, I assume that someone can have a module name larger than four characters. Make sense? Or Did I missed something?
Sure, names can and are already larger than four characters (e.g. "amdgpu"). But the driver matching uses the first four characters of the name. The length of matching the driver name doesn't need to be lengthened as of yet.
Kernel module name, again, is another thing entirely.
Signature · 5 lines
Petri Latvala _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/2 Add support for forcing specific module
In reply to Rodrigo SiqueiraThis commit adds a new option for forcing the use of a specific module indicated via command line. The force command expects a module name which will be used on the target test (changing the standard behavior).
--- lib/drmtest.c | 33 +++++++++++++++++++++++++++------ lib/igt_core.c | 23 +++++++++++++++++++++++ lib/igt_core.h | 4 ++++ scripts/run-tests.sh | 4 +++- 4 files changed, 57 insertions(+), 7 deletions(-)
lib/drmtest.c +27 -6
@@ -247,6 +247,20 @@ static int __open_device(const char *base, int offset, unsigned int chipset)if (chipset &DRIVER_AMDGPU &&is_amd_device (fd))return fd; +// Force options +if (get_target_module () && chipset ==DRIVER_ANY ) { +if (__is_device (fd,get_target_module ())) { +return fd; + } +else { +memset (name,0 ,sizeof (name)); +__get_drm_device_name (fd, name); +close (fd); +igt_kmod_unload (name,0 ); +return -1 ; + } + } +/* Only VGEM-specific tests should be run on VGEM */ if (chipset ==DRIVER_ANY && !is_vgem_device (fd))return fd;
@@ -260,6 +274,7 @@ static int __open_device(const char *base, int offset, unsigned int chipset)static int __open_driver (const char *base,int offset,unsigned int chipset) {static pthread_mutex_t mutex =PTHREAD_MUTEX_INITIALIZER ; +const char * target; static const struct module { unsigned int bit; const char *module;
@@ -278,13 +293,19 @@ static int __open_driver(const char *base, int offset, unsigned int chipset)if (fd !=-1 )return fd; + target =get_target_module (); +pthread_mutex_lock (&mutex); -for (const struct module *m = modules; m->module; m++) { -if (chipset & m->bit) { -if (m->modprobe) - m->modprobe (m->module); -else -modprobe (m->module); +if (target) { +modprobe (target); + }else { +for (const struct module *m = modules; m->module; m++) { +if (chipset & m->bit) { +if (m->modprobe) + m->modprobe (m->module); +else +modprobe (m->module); + } } }pthread_mutex_unlock (&mutex);
lib/igt_core.c +23 -0
@@ -286,6 +286,7 @@ enum {OPT_DESCRIPTION ,OPT_DEBUG ,OPT_INTERACTIVE_DEBUG , +OPT_FORCE_MODULE , OPT_HELP = 'h' };
@@ -303,6 +304,23 @@ static pthread_mutex_t log_buffer_mutex = PTHREAD_MUTEX_INITIALIZER;GKeyFile *igt_key_file;#endif +char *target_module =NULL ; + +void set_target_module (char *target) +{ +if (!target) +igt_warn ("No module specified, keep default behaviour" ); + + target_module = target; +} + +const char *get_target_module (void ) +{ +if (target_module) +return target_module; +return NULL ; +} +char *igt_frame_dump_path;const char *igt_test_name (void )
@@ -555,6 +573,7 @@ static void print_usage(const char *help_str, bool output_on_stderr)" --debug[=log-domain]\n" " --interactive-debug[=domain]\n" " --help-description\n" +" --force-module <module>\n" " --help\n");if (help_str)fprintf (f,"%s\n" , help_str);
@@ -666,6 +685,7 @@ static int common_init(int *argc, char **argv, {"debug", optional_argument, 0, OPT_DEBUG}, {"interactive-debug", optional_argument, 0, OPT_INTERACTIVE_DEBUG}, {"help", 0, 0, OPT_HELP}, + {"force-module", required_argument, 0, OPT_FORCE_MODULE}, {0, 0, 0, 0} };char *short_opts;
@@ -763,6 +783,9 @@ static int common_init(int *argc, char **argv,print_test_description (); ret =-1 ; goto out; +case OPT_FORCE_MODULE : +set_target_module (strdup (optarg)); +break ;case OPT_HELP :print_usage (help_str, false); ret =-1 ;
lib/igt_core.h +4 -0
@@ -50,6 +50,10 @@ extern const char* __igt_test_description __attribute__((weak));extern bool __igt_plain_output;extern char *igt_frame_dump_path; +extern char *target_module; +void set_target_module (char *target); +const char *get_target_module (void ); +/** * IGT_TEST_DESCRIPTION: * @str: description string
scripts/run-tests.sh +3 -1
@@ -80,6 +80,7 @@ function print_help { echo " (can be used more than once)" echo " -T <filename> run tests listed in testlist" echo " (overrides -t and -x)" + echo " -f <module> force specific module load" echo " -v enable verbose mode" echo " -x <regex> exclude tests that match the regular expression" echo " (can be used more than once)"
@@ -91,7 +92,7 @@ function print_help { echo "Useful patterns for test filtering are described in the API documentation." } -while getopts ":dhlr:st:T:vx:Rn" opt; do +while getopts ":dhlr:st:T:m:vx:Rn" opt; do case $opt in d) download_piglit; exit ;; h) print_help; exit ;;
@@ -100,6 +101,7 @@ while getopts ":dhlr:st:T:vx:Rn" opt; do s) SUMMARY="html" ;; t) FILTER="$FILTER -t $OPTARG" ;; T) FILTER="$FILTER --test-list $OPTARG" ;; + f) FILTER="$FILTER --force-module $OPTARG" ;; v) VERBOSE="-v" ;; x) EXCLUDE="$EXCLUDE -x $OPTARG" ;; R) RESUME="true" ;;
Signature · 6 lines
2.18.0 _______________________________________________ igt-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/igt-dev
Re: [Intel-gfx] [PATCH i-g-t 2/2] Add support for forcing specific module
Quoted 24 lines · Rodrigo Siqueira
This commit adds a new option for forcing the use of a specific module indicated via command line. The force command expects a module name which will be used on the target test (changing the standard behavior).
Signed-off-by Rodrigo Siqueira <[email protected]>--- lib/drmtest.c | 33 +++++++++++++++++++++++++++------ lib/igt_core.c | 23 +++++++++++++++++++++++ lib/igt_core.h | 4 ++++ scripts/run-tests.sh | 4 +++- 4 files changed, 57 insertions(+), 7 deletions(-)lib/drmtest.c +4
-0@@ -247,6 +247,20 @@ static int __open_device(const char *base, int offset, unsigned int chipset)if (chipset &DRIVER_AMDGPU &&is_amd_device (fd))return fd; +// Force options +if (get_target_module () && chipset ==DRIVER_ANY ) { +if (__is_device (fd,get_target_module ())) { +return fd;
lib/drmtest.c static int __open_device(const char *base, int offset, unsigned int chipset) · line 263
Quoted 10 lines
lib/drmtest.c +10
-0+ } +else { +memset (name,0 ,sizeof (name)); +__get_drm_device_name (fd, name); +close (fd); +igt_kmod_unload (name,0 ); +return -1 ; + } + } +
You're unloading modules (which you shouldn't need to do here anyway), and even worse, unloading them if the first loop iteration doesn't land on the forced name.
This chunk needs to be before the matches for the known bits for DRIVER_ANY or you'll just get one of them when forcing a custom name.
Quoted 17 lines
lib/drmtest.c +3
-0/* Only VGEM-specific tests should be run on VGEM */ if (chipset ==DRIVER_ANY && !is_vgem_device (fd))return fd;@@ -260,6 +274,7 @@ static int __open_device(const char *base, int offset, unsigned int chipset)static int __open_driver (const char *base,int offset,unsigned int chipset) {static pthread_mutex_t mutex =PTHREAD_MUTEX_INITIALIZER ; +const char * target; static const struct module { unsigned int bit; const char *module;@@ -278,13 +293,19 @@ static int __open_driver(const char *base, int offset, unsigned int chipset)if (fd !=-1 )return fd; + target =get_target_module (); +
lib/drmtest.c static int __open_driver(const char *base, int offset, unsigned int chipset) · line 301
Quoted 10 lines
lib/drmtest.c +3
-6pthread_mutex_lock (&mutex); -for (const struct module *m = modules; m->module; m++) { -if (chipset & m->bit) { -if (m->modprobe) - m->modprobe (m->module); -else -modprobe (m->module); +if (target) { +modprobe (target); + }else {
The name of the driver (as returned by DRM_IOCTL_VERSION) doesn't have to match the kernel module filename. Case in point: virtio-gpu.ko is "virtio_gpu".
I would suggest leaving the kernel module loading plans later for now and get the DRIVER_ANY match forcing in place. Or even forget kernel module loading entirely. If the user is able to set a parameter to a module name, the user can load the module themselves before running tests. It won't get unloaded if IGT doesn't have a test that does it, and it won't have such a test unless it knows about the driver/module. And at that point, IGT will know how to load that module.
Quoted 109 lines
lib/drmtest.c +7
-0+for (const struct module *m = modules; m->module; m++) { +if (chipset & m->bit) { +if (m->modprobe) + m->modprobe (m->module); +else +modprobe (m->module); + } } }pthread_mutex_unlock (&mutex);lib/igt_core.c +23
-0@@ -286,6 +286,7 @@ enum {OPT_DESCRIPTION ,OPT_DEBUG ,OPT_INTERACTIVE_DEBUG , +OPT_FORCE_MODULE , OPT_HELP = 'h' };@@ -303,6 +304,23 @@ static pthread_mutex_t log_buffer_mutex = PTHREAD_MUTEX_INITIALIZER;GKeyFile *igt_key_file;#endif +char *target_module =NULL ; + +void set_target_module (char *target) +{ +if (!target) +igt_warn ("No module specified, keep default behaviour" ); + + target_module = target; +} + +const char *get_target_module (void ) +{ +if (target_module) +return target_module; +return NULL ; +} +char *igt_frame_dump_path;const char *igt_test_name (void )@@ -555,6 +573,7 @@ static void print_usage(const char *help_str, bool output_on_stderr)" --debug[=log-domain]\n" " --interactive-debug[=domain]\n" " --help-description\n" +" --force-module <module>\n" " --help\n");if (help_str)fprintf (f,"%s\n" , help_str);@@ -666,6 +685,7 @@ static int common_init(int *argc, char **argv, {"debug", optional_argument, 0, OPT_DEBUG}, {"interactive-debug", optional_argument, 0, OPT_INTERACTIVE_DEBUG}, {"help", 0, 0, OPT_HELP}, + {"force-module", required_argument, 0, OPT_FORCE_MODULE}, {0, 0, 0, 0} };char *short_opts;@@ -763,6 +783,9 @@ static int common_init(int *argc, char **argv,print_test_description (); ret =-1 ; goto out; +case OPT_FORCE_MODULE : +set_target_module (strdup (optarg)); +break ;case OPT_HELP :print_usage (help_str, false); ret =-1 ;lib/igt_core.h +4
-0@@ -50,6 +50,10 @@ extern const char* __igt_test_description __attribute__((weak));extern bool __igt_plain_output;extern char *igt_frame_dump_path; +extern char *target_module; +void set_target_module (char *target); +const char *get_target_module (void ); +/** * IGT_TEST_DESCRIPTION: * @str: description string scripts/run-tests.sh +2
-1@@ -80,6 +80,7 @@ function print_help { echo " (can be used more than once)" echo " -T <filename> run tests listed in testlist" echo " (overrides -t and -x)" + echo " -f <module> force specific module load" echo " -v enable verbose mode" echo " -x <regex> exclude tests that match the regular expression" echo " (can be used more than once)"@@ -91,7 +92,7 @@ function print_help { echo "Useful patterns for test filtering are described in the API documentation." } -while getopts ":dhlr:st:T:vx:Rn" opt; do +while getopts ":dhlr:st:T:m:vx:Rn" opt; do case $opt in d) download_piglit; exit ;; h) print_help; exit ;;
scripts/run-tests.sh while getopts ":dhlr:st:T:vx:Rn" opt; do · line 104
Quoted 4 lines
scripts/run-tests.sh +1
-0@@ -100,6 +101,7 @@ while getopts ":dhlr:st:T:vx:Rn" opt; do s) SUMMARY="html" ;; t) FILTER="$FILTER -t $OPTARG" ;; T) FILTER="$FILTER --test-list $OPTARG" ;; + f) FILTER="$FILTER --force-module $OPTARG" ;;
This command line is given to piglit, not the IGT tests.
Furthermore, run-tests.sh is not required to run tests. One can execute piglit or igt_runner directly.
Rather, consider having the force driver name passed via environment variables.
Signature · 5 lines
Petri Latvala _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [igt-dev] [PATCH i-g-t 2/2] Add support for forcing specific module
Quoted 41 lines · Petri Latvala
Quoted 24 lines · Rodrigo Siqueira
This commit adds a new option for forcing the use of a specific module indicated via command line. The force command expects a module name which will be used on the target test (changing the standard behavior).
Signed-off-by Rodrigo Siqueira <[email protected]>--- lib/drmtest.c | 33 +++++++++++++++++++++++++++------ lib/igt_core.c | 23 +++++++++++++++++++++++ lib/igt_core.h | 4 ++++ scripts/run-tests.sh | 4 +++- 4 files changed, 57 insertions(+), 7 deletions(-)lib/drmtest.c +4
-0@@ -247,6 +247,20 @@ static int __open_device(const char *base, int offset, unsigned int chipset)if (chipset &DRIVER_AMDGPU &&is_amd_device (fd))return fd; +// Force options +if (get_target_module () && chipset ==DRIVER_ANY ) { +if (__is_device (fd,get_target_module ())) { +return fd;lib/drmtest.c static int __open_device(const char *base, int offset, unsigned int chipset) · line 263
Quoted 10 lines
lib/drmtest.c +10
-0+ } +else { +memset (name,0 ,sizeof (name)); +__get_drm_device_name (fd, name); +close (fd); +igt_kmod_unload (name,0 ); +return -1 ; + } + } +You're unloading modules (which you shouldn't need to do here anyway), and even worse, unloading them if the first loop iteration doesn't land on the forced name.
Quoted 2 lines
This chunk needs to be before the matches for the known bits for
DRIVER_ANYor you'll just get one of them when forcing a custom name.
Hi,
I think I did not fully understand your point; I'm a little bit confused with this part:
"[..] or you'll just get one of them when forcing a custom name."
Do we want more than one?
Quoted 34 lines
Quoted 18 lines
lib/drmtest.c +3
-0/* Only VGEM-specific tests should be run on VGEM */ if (chipset ==DRIVER_ANY && !is_vgem_device (fd))return fd;@@ -260,6 +274,7 @@ static int __open_device(const char *base, int offset, unsigned int chipset)static int __open_driver (const char *base,int offset,unsigned int chipset) {static pthread_mutex_t mutex =PTHREAD_MUTEX_INITIALIZER ; +const char * target; static const struct module { unsigned int bit; const char *module;@@ -278,13 +293,19 @@ static int __open_driver(const char *base, int offset, unsigned int chipset)if (fd !=-1 )return fd; + target =get_target_module (); +lib/drmtest.c static int __open_driver(const char *base, int offset, unsigned int chipset) · line 301
Quoted 10 lines
lib/drmtest.c +3
-6pthread_mutex_lock (&mutex); -for (const struct module *m = modules; m->module; m++) { -if (chipset & m->bit) { -if (m->modprobe) - m->modprobe (m->module); -else -modprobe (m->module); +if (target) { +modprobe (target); + }else {The name of the driver (as returned by
DRM_IOCTL_VERSION) doesn't have to match the kernel module filename. Case in point: virtio-gpu.ko is "virtio_gpu".
Quoted 8 lines
I would suggest leaving the kernel module loading plans later for now and get the
DRIVER_ANYmatch forcing in place. Or even forget kernel module loading entirely. If the user is able to set a parameter to a module name, the user can load the module themselves before running tests. It won't get unloaded if IGT doesn't have a test that does it, and it won't have such a test unless it knows about the driver/module. And at that point, IGT will know how to load that module.
Basically, leave the modprobe part for the user, Right?
Quoted 119 lines
Quoted 111 lines
lib/drmtest.c +7
-0+for (const struct module *m = modules; m->module; m++) { +if (chipset & m->bit) { +if (m->modprobe) + m->modprobe (m->module); +else +modprobe (m->module); + } } }pthread_mutex_unlock (&mutex);lib/igt_core.c +23
-0@@ -286,6 +286,7 @@ enum {OPT_DESCRIPTION ,OPT_DEBUG ,OPT_INTERACTIVE_DEBUG , +OPT_FORCE_MODULE , OPT_HELP = 'h' };@@ -303,6 +304,23 @@ static pthread_mutex_t log_buffer_mutex = PTHREAD_MUTEX_INITIALIZER;GKeyFile *igt_key_file;#endif +char *target_module =NULL ; + +void set_target_module (char *target) +{ +if (!target) +igt_warn ("No module specified, keep default behaviour" ); + + target_module = target; +} + +const char *get_target_module (void ) +{ +if (target_module) +return target_module; +return NULL ; +} +char *igt_frame_dump_path;const char *igt_test_name (void )@@ -555,6 +573,7 @@ static void print_usage(const char *help_str, bool output_on_stderr)" --debug[=log-domain]\n" " --interactive-debug[=domain]\n" " --help-description\n" +" --force-module <module>\n" " --help\n");if (help_str)fprintf (f,"%s\n" , help_str);@@ -666,6 +685,7 @@ static int common_init(int *argc, char **argv, {"debug", optional_argument, 0, OPT_DEBUG}, {"interactive-debug", optional_argument, 0, OPT_INTERACTIVE_DEBUG}, {"help", 0, 0, OPT_HELP}, + {"force-module", required_argument, 0, OPT_FORCE_MODULE}, {0, 0, 0, 0} };char *short_opts;@@ -763,6 +783,9 @@ static int common_init(int *argc, char **argv,print_test_description (); ret =-1 ; goto out; +case OPT_FORCE_MODULE : +set_target_module (strdup (optarg)); +break ;case OPT_HELP :print_usage (help_str, false); ret =-1 ;lib/igt_core.h +4
-0@@ -50,6 +50,10 @@ extern const char* __igt_test_description __attribute__((weak));extern bool __igt_plain_output;extern char *igt_frame_dump_path; +extern char *target_module; +void set_target_module (char *target); +const char *get_target_module (void ); +/** * IGT_TEST_DESCRIPTION: * @str: description string scripts/run-tests.sh +2
-1@@ -80,6 +80,7 @@ function print_help { echo " (can be used more than once)" echo " -T <filename> run tests listed in testlist" echo " (overrides -t and -x)" + echo " -f <module> force specific module load" echo " -v enable verbose mode" echo " -x <regex> exclude tests that match the regular expression" echo " (can be used more than once)"@@ -91,7 +92,7 @@ function print_help { echo "Useful patterns for test filtering are described in the API documentation." } -while getopts ":dhlr:st:T:vx:Rn" opt; do +while getopts ":dhlr:st:T:m:vx:Rn" opt; do case $opt in d) download_piglit; exit ;; h) print_help; exit ;;scripts/run-tests.sh while getopts ":dhlr:st:T:vx:Rn" opt; do · line 104
Quoted 4 lines
scripts/run-tests.sh +1
-0@@ -100,6 +101,7 @@ while getopts ":dhlr:st:T:vx:Rn" opt; do s) SUMMARY="html" ;; t) FILTER="$FILTER -t $OPTARG" ;; T) FILTER="$FILTER --test-list $OPTARG" ;; + f) FILTER="$FILTER --force-module $OPTARG" ;;This command line is given to piglit, not the IGT tests.
Quoted 5 lines
Furthermore,
run-tests.shis not required to run tests. One can execute piglit origt_runnerdirectly.Rather, consider having the force driver name passed via environment variables.
Do you mean, something like this:
kms_flip --run-subtest basic-plain-flip MODULE=vkms
Thanks a lot for the review!
Best Regards
Quoted 3 lines
Signature · 2 lines
-- Petri Latvala
Signature · 9 lines
Rodrigo Siqueira http://siqueira.tech Graduate Student Department of Computer Science University of São Paulo _______________________________________________ igt-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/igt-dev
Re: [igt-dev] [PATCH i-g-t 2/2] Add support for forcing specific module
Quoted 44 lines · Rodrigo Siqueira
Quoted 41 lines · Petri Latvala
Quoted 24 lines · Rodrigo Siqueira
This commit adds a new option for forcing the use of a specific module indicated via command line. The force command expects a module name which will be used on the target test (changing the standard behavior).
Signed-off-by Rodrigo Siqueira <[email protected]>--- lib/drmtest.c | 33 +++++++++++++++++++++++++++------ lib/igt_core.c | 23 +++++++++++++++++++++++ lib/igt_core.h | 4 ++++ scripts/run-tests.sh | 4 +++- 4 files changed, 57 insertions(+), 7 deletions(-)lib/drmtest.c +4
-0@@ -247,6 +247,20 @@ static int __open_device(const char *base, int offset, unsigned int chipset)if (chipset &DRIVER_AMDGPU &&is_amd_device (fd))return fd; +// Force options +if (get_target_module () && chipset ==DRIVER_ANY ) { +if (__is_device (fd,get_target_module ())) { +return fd;lib/drmtest.c static int __open_device(const char *base, int offset, unsigned int chipset) · line 263
Quoted 10 lines
lib/drmtest.c +10
-0+ } +else { +memset (name,0 ,sizeof (name)); +__get_drm_device_name (fd, name); +close (fd); +igt_kmod_unload (name,0 ); +return -1 ; + } + } +You're unloading modules (which you shouldn't need to do here anyway), and even worse, unloading them if the first loop iteration doesn't land on the forced name.
Quoted 2 lines
This chunk needs to be before the matches for the known bits for
DRIVER_ANYor you'll just get one of them when forcing a custom name.
Quoted 8 lines
Hi,
I think I did not fully understand your point; I'm a little bit confused with this part:
"[..] or you'll just get one of them when forcing a custom name."
Do we want more than one?
I mean that when a test calls drm_open_driver(DRIVER_ANY), it will happily open /dev/dri/card0 and use that if it's, say, an i915 device. Even though you enabled the force option to "vkms".
Quoted 44 lines
Quoted 35 lines
Quoted 19 lines
lib/drmtest.c +3
-0/* Only VGEM-specific tests should be run on VGEM */ if (chipset ==DRIVER_ANY && !is_vgem_device (fd))return fd;@@ -260,6 +274,7 @@ static int __open_device(const char *base, int offset, unsigned int chipset)static int __open_driver (const char *base,int offset,unsigned int chipset) {static pthread_mutex_t mutex =PTHREAD_MUTEX_INITIALIZER ; +const char * target; static const struct module { unsigned int bit; const char *module;@@ -278,13 +293,19 @@ static int __open_driver(const char *base, int offset, unsigned int chipset)if (fd !=-1 )return fd; + target =get_target_module (); +lib/drmtest.c static int __open_driver(const char *base, int offset, unsigned int chipset) · line 301
Quoted 10 lines
lib/drmtest.c +3
-6pthread_mutex_lock (&mutex); -for (const struct module *m = modules; m->module; m++) { -if (chipset & m->bit) { -if (m->modprobe) - m->modprobe (m->module); -else -modprobe (m->module); +if (target) { +modprobe (target); + }else {The name of the driver (as returned by
DRM_IOCTL_VERSION) doesn't have to match the kernel module filename. Case in point: virtio-gpu.ko is "virtio_gpu".Quoted 8 lines
I would suggest leaving the kernel module loading plans later for now and get the
DRIVER_ANYmatch forcing in place. Or even forget kernel module loading entirely. If the user is able to set a parameter to a module name, the user can load the module themselves before running tests. It won't get unloaded if IGT doesn't have a test that does it, and it won't have such a test unless it knows about the driver/module. And at that point, IGT will know how to load that module.
Quoted 1 line
Basically, leave the modprobe part for the user, Right?
Exactly.
Quoted 123 lines
Quoted 123 lines
Quoted 112 lines
lib/drmtest.c +7
-0+for (const struct module *m = modules; m->module; m++) { +if (chipset & m->bit) { +if (m->modprobe) + m->modprobe (m->module); +else +modprobe (m->module); + } } }pthread_mutex_unlock (&mutex);lib/igt_core.c +23
-0@@ -286,6 +286,7 @@ enum {OPT_DESCRIPTION ,OPT_DEBUG ,OPT_INTERACTIVE_DEBUG , +OPT_FORCE_MODULE , OPT_HELP = 'h' };@@ -303,6 +304,23 @@ static pthread_mutex_t log_buffer_mutex = PTHREAD_MUTEX_INITIALIZER;GKeyFile *igt_key_file;#endif +char *target_module =NULL ; + +void set_target_module (char *target) +{ +if (!target) +igt_warn ("No module specified, keep default behaviour" ); + + target_module = target; +} + +const char *get_target_module (void ) +{ +if (target_module) +return target_module; +return NULL ; +} +char *igt_frame_dump_path;const char *igt_test_name (void )@@ -555,6 +573,7 @@ static void print_usage(const char *help_str, bool output_on_stderr)" --debug[=log-domain]\n" " --interactive-debug[=domain]\n" " --help-description\n" +" --force-module <module>\n" " --help\n");if (help_str)fprintf (f,"%s\n" , help_str);@@ -666,6 +685,7 @@ static int common_init(int *argc, char **argv, {"debug", optional_argument, 0, OPT_DEBUG}, {"interactive-debug", optional_argument, 0, OPT_INTERACTIVE_DEBUG}, {"help", 0, 0, OPT_HELP}, + {"force-module", required_argument, 0, OPT_FORCE_MODULE}, {0, 0, 0, 0} };char *short_opts;@@ -763,6 +783,9 @@ static int common_init(int *argc, char **argv,print_test_description (); ret =-1 ; goto out; +case OPT_FORCE_MODULE : +set_target_module (strdup (optarg)); +break ;case OPT_HELP :print_usage (help_str, false); ret =-1 ;lib/igt_core.h +4
-0@@ -50,6 +50,10 @@ extern const char* __igt_test_description __attribute__((weak));extern bool __igt_plain_output;extern char *igt_frame_dump_path; +extern char *target_module; +void set_target_module (char *target); +const char *get_target_module (void ); +/** * IGT_TEST_DESCRIPTION: * @str: description string scripts/run-tests.sh +2
-1@@ -80,6 +80,7 @@ function print_help { echo " (can be used more than once)" echo " -T <filename> run tests listed in testlist" echo " (overrides -t and -x)" + echo " -f <module> force specific module load" echo " -v enable verbose mode" echo " -x <regex> exclude tests that match the regular expression" echo " (can be used more than once)"@@ -91,7 +92,7 @@ function print_help { echo "Useful patterns for test filtering are described in the API documentation." } -while getopts ":dhlr:st:T:vx:Rn" opt; do +while getopts ":dhlr:st:T:m:vx:Rn" opt; do case $opt in d) download_piglit; exit ;; h) print_help; exit ;;scripts/run-tests.sh while getopts ":dhlr:st:T:vx:Rn" opt; do · line 104
Quoted 4 lines
scripts/run-tests.sh +1
-0@@ -100,6 +101,7 @@ while getopts ":dhlr:st:T:vx:Rn" opt; do s) SUMMARY="html" ;; t) FILTER="$FILTER -t $OPTARG" ;; T) FILTER="$FILTER --test-list $OPTARG" ;; + f) FILTER="$FILTER --force-module $OPTARG" ;;This command line is given to piglit, not the IGT tests.
Furthermore,
run-tests.shis not required to run tests. One can execute piglit origt_runnerdirectly.
Quoted 6 lines
Quoted 2 lines
Rather, consider having the force driver name passed via environment variables.
Do you mean, something like this:
kms_flip--run-subtest basic-plain-flip MODULE=vkms
MODULE=vkms kms_flip --run-subtest basic-plain-flip
Or the more usual way of running tests:
IGT_TEST_ROOT=builddir/tests MODULE=vkms piglit run igt -t igt@kms_flip@basic-plain-flip results-directory
or
MODULE=vkms builddir/runner/igt_runner -t igt@kms_flip@basic-plain-flip builddir/tests results-directory
Signature · 5 lines
Petri Latvala _______________________________________________ igt-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/igt-dev
Patchwork
CI[igt-dev] ✓ Fi.CI.BAT: success for Add support to force specific module load (rev2)
In reply to Rodrigo Siqueira== Series Details ==
== Series Details ==
Series: Add support to force specific module load (rev2)
URL : https://patchwork.freedesktop.org/series/43912/
State : success
== Summary ==
= CI Bug Log - changes from CI_DRM_4445 -> IGTPW_1543 =
== Summary - SUCCESS ==
No regressions found.
External URL: https://patchwork.freedesktop.org/api/1.0/series/43912/revisions/2/mbox/
== Known issues ==
Here are the changes found in IGTPW_1543 that come from known issues:
=== IGT changes ===
==== Issues hit ====
igt@kms_frontbuffer_tracking@basic:
fi-bxt-j4205: PASS -> FAIL (fdo#103167)
fi-skl-6700hq: PASS -> FAIL (fdo#103167)
fi-bxt-dsi: PASS -> FAIL (fdo#103167)
fi-skl-6700k2: PASS -> FAIL (fdo#103167)
fi-kbl-7560u: PASS -> FAIL (fdo#103167)
fi-skl-6600u: PASS -> FAIL (fdo#103167)
fi-kbl-r: PASS -> FAIL (fdo#103167)
{fi-cfl-8109u}: NOTRUN -> FAIL (fdo#103167)
fi-skl-6260u: PASS -> FAIL (fdo#103167)
fi-skl-6770hq: PASS -> FAIL (fdo#103167)
fi-cfl-guc: PASS -> FAIL (fdo#103167)
fi-cfl-s3: PASS -> FAIL (fdo#103167)
fi-whl-u: PASS -> FAIL (fdo#103167)
fi-cfl-8700k: PASS -> FAIL (fdo#103167)
fi-kbl-7500u: PASS -> FAIL (fdo#103167)
fi-kbl-7567u: PASS -> FAIL (fdo#103167)
fi-skl-guc: PASS -> FAIL (fdo#103167)
fi-glk-j4005: PASS -> FAIL (fdo#103167)
fi-glk-dsi: PASS -> FAIL (fdo#103167)
fi-skl-gvtdvm: PASS -> FAIL (fdo#103167)
igt@kms_pipe_crc_basic@suspend-read-crc-pipe-b:
fi-snb-2520m: PASS -> INCOMPLETE (fdo#103713)
==== Possible fixes ====
igt@gem_exec_suspend@basic-s3:
{fi-cfl-8109u}: INCOMPLETE -> PASS {name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).
fdo#103167 https://bugs.freedesktop.org/show_bug.cgi?id=103167
fdo#103713 https://bugs.freedesktop.org/show_bug.cgi?id=103713
== Participating hosts (47 -> 41) ==
Missing (6): fi-ilk-m540 fi-hsw-4200u fi-byt-squawks fi-bsw-cyan fi-ctg-p8600 fi-kbl-8809g
== Build changes ==
* IGT: IGT_4540 -> IGTPW_1543
CI_DRM_4445: 366234e5605ce58f732d2aeb83c713dde6b7c85c @ git://anongit.freedesktop.org/gfx-ci/linux
IGTPW_1543: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_1543/
IGT_4540: 78071c2fa53db2f04b8eddc6e6118be4fbc5c2fe @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_1543/issues.html
_______________________________________________
igt-dev mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/igt-dev
Patchwork
CI[igt-dev] ✗ Fi.CI.IGT: failure for Add support to force specific module load (rev2)
In reply to Rodrigo Siqueira== Series Details ==
== Series Details ==
Series: Add support to force specific module load (rev2)
URL : https://patchwork.freedesktop.org/series/43912/
State : failure
== Summary ==
= CI Bug Log - changes from IGT_4540_full -> IGTPW_1543_full =
== Summary - FAILURE ==
Serious unknown changes coming with IGTPW_1543_full absolutely need to be verified manually.
If you think the reported changes have nothing to do with the changes introduced in IGTPW_1543_full, please notify your bug team to allow them to document this new failure mode, which will reduce false positives in CI.
External URL: https://patchwork.freedesktop.org/api/1.0/series/43912/revisions/2/mbox/
== Possible new issues ==
Here are the unknown changes that may have been introduced in IGTPW_1543_full:
=== IGT changes ===
==== Possible regressions ====
igt@kms_draw_crc@draw-method-xrgb8888-render-untiled:
shard-glk: PASS -> FAIL +5
shard-kbl: PASS -> FAIL +1
igt@kms_draw_crc@draw-method-xrgb8888-render-xtiled:
shard-apl: PASS -> FAIL +1
==== Warnings ====
igt@gem_exec_schedule@deep-bsd1:
shard-kbl: PASS -> SKIP +1
igt@gem_mocs_settings@mocs-rc6-ctx-render:
shard-kbl: SKIP -> PASS== Known issues ==
Here are the changes found in IGTPW_1543_full that come from known issues:
=== IGT changes ===
==== Issues hit ====
igt@drv_selftest@live_gtt:
shard-kbl: PASS -> INCOMPLETE (fdo#103665, fdo#107127)
igt@gem_busy@busy-bsd2:
shard-snb: SKIP -> INCOMPLETE (fdo#105411)
igt@gem_ppgtt@blt-vs-render-ctx0:
shard-kbl: PASS -> INCOMPLETE (fdo#106023, fdo#103665)
igt@kms_available_modes_crc@available_mode_test_crc:
shard-snb: PASS -> FAIL (fdo#106641)
igt@kms_draw_crc@draw-method-xrgb2101010-render-untiled:
shard-apl: PASS -> FAIL (fdo#103184) +2
igt@kms_draw_crc@draw-method-xrgb2101010-render-xtiled:
shard-kbl: PASS -> FAIL (fdo#103184) +2
igt@kms_draw_crc@draw-method-xrgb8888-render-ytiled:
shard-apl: PASS -> FAIL (fdo#103232)
shard-glk: PASS -> FAIL (fdo#103232)
shard-kbl: PASS -> FAIL (fdo#103232)
igt@kms_flip@2x-flip-vs-expired-vblank:
shard-glk: PASS -> FAIL (fdo#105189)
igt@kms_flip@flip-vs-wf_vblank-interruptible:
shard-glk: PASS -> FAIL (fdo#103928)
igt@kms_flip_tiling@flip-x-tiled:
shard-glk: PASS -> FAIL (fdo#103822) +1
igt@kms_frontbuffer_tracking@fbc-1p-primscrn-cur-indfb-draw-render:
shard-apl: PASS -> FAIL (fdo#103167) +5
igt@kms_frontbuffer_tracking@fbc-1p-primscrn-spr-indfb-draw-render:
shard-kbl: PASS -> FAIL (fdo#103167) +5
igt@kms_frontbuffer_tracking@fbc-2p-pri-indfb-multidraw:
shard-glk: PASS -> FAIL (fdo#103167) +10
igt@kms_plane_lowres@pipe-a-tiling-x:
shard-glk: PASS -> FAIL (fdo#103166)
igt@kms_rotation_crc@primary-rotation-180:
shard-snb: PASS -> FAIL (fdo#103925)
igt@kms_setmode@basic:
shard-apl: PASS -> FAIL (fdo#99912)
==== Possible fixes ====
igt@kms_flip@2x-plain-flip-ts-check-interruptible:
shard-glk: FAIL (fdo#100368) -> PASS
igt@kms_flip@flip-vs-expired-vblank-interruptible:
shard-glk: FAIL (fdo#105363, fdo#102887) -> PASS
igt@kms_flip_tiling@flip-to-y-tiled:
shard-glk: FAIL (fdo#103822) -> PASS
igt@kms_frontbuffer_tracking@fbc-1p-indfb-fliptrack:
shard-snb: INCOMPLETE (fdo#105411) -> PASS
igt@testdisplay:
shard-glk: INCOMPLETE (k.org#198133, fdo#103359) -> PASS
==== Warnings ====
igt@drv_selftest@live_gtt:
shard-apl: FAIL (fdo#105347, fdo#107127) -> INCOMPLETE (fdo#103927, fdo#107127) fdo#100368 https://bugs.freedesktop.org/show_bug.cgi?id=100368
fdo#102887 https://bugs.freedesktop.org/show_bug.cgi?id=102887
fdo#103166 https://bugs.freedesktop.org/show_bug.cgi?id=103166
fdo#103167 https://bugs.freedesktop.org/show_bug.cgi?id=103167
fdo#103184 https://bugs.freedesktop.org/show_bug.cgi?id=103184
fdo#103232 https://bugs.freedesktop.org/show_bug.cgi?id=103232
fdo#103359 https://bugs.freedesktop.org/show_bug.cgi?id=103359
fdo#103665 https://bugs.freedesktop.org/show_bug.cgi?id=103665
fdo#103822 https://bugs.freedesktop.org/show_bug.cgi?id=103822
fdo#103925 https://bugs.freedesktop.org/show_bug.cgi?id=103925
fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927
fdo#103928 https://bugs.freedesktop.org/show_bug.cgi?id=103928
fdo#105189 https://bugs.freedesktop.org/show_bug.cgi?id=105189
fdo#105347 https://bugs.freedesktop.org/show_bug.cgi?id=105347
fdo#105363 https://bugs.freedesktop.org/show_bug.cgi?id=105363
fdo#105411 https://bugs.freedesktop.org/show_bug.cgi?id=105411
fdo#106023 https://bugs.freedesktop.org/show_bug.cgi?id=106023
fdo#106641 https://bugs.freedesktop.org/show_bug.cgi?id=106641
fdo#107127 https://bugs.freedesktop.org/show_bug.cgi?id=107127
fdo#99912 https://bugs.freedesktop.org/show_bug.cgi?id=99912
k.org#198133 https://bugzilla.kernel.org/show_bug.cgi?id=198133
== Participating hosts (5 -> 5) ==
No changes in participating hosts
== Build changes ==
* IGT: IGT_4540 -> IGTPW_1543
* Linux: CI_DRM_4443 -> CI_DRM_4445
CI_DRM_4443: 5c43ea095bbd1469a9c767529537ddf0434acc60 @ git://anongit.freedesktop.org/gfx-ci/linux
CI_DRM_4445: 366234e5605ce58f732d2aeb83c713dde6b7c85c @ git://anongit.freedesktop.org/gfx-ci/linux
IGTPW_1543: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_1543/
IGT_4540: 78071c2fa53db2f04b8eddc6e6118be4fbc5c2fe @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_1543/shards.html
_______________________________________________
igt-dev mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/igt-dev
Hi,
I would like to know if there is any improvement that I can do for this patchset.
Thanks
Quoted 25 lines · Rodrigo Siqueira
This patchset adds support for forcing a specific module to be loaded via command line. This feature can bring benefits for developers that need to create a new module since they can use IGT as a base test to guide their development (Test-driven development - TDD ). The first patch, expand the size of the module name accepted by IGT to make possible that modules with larger names can be loaded. The second patch has the force option implementation.
The discussion about this feature started with an RFC named "[PATCH i-g-t] [RFC] Add support to force specific module load", link:
https://www.spinics.net/lists/intel-gfx/msg166764.htmlRodrigo Siqueira (2):
Increase the string size for a module name
Add support for forcing specific modulelib/drmtest.c | 39 +++++++++++++++++++++++++++++++-------- lib/igt_core.c | 23 +++++++++++++++++++++++ lib/igt_core.h | 4 ++++ scripts/run-tests.sh | 4 +++- 4 files changed, 61 insertions(+), 9 deletions(-)Signature · 1 line
2.18.0
Signature · 9 lines
Rodrigo Siqueira http://siqueira.tech Graduate Student Department of Computer Science University of São Paulo _______________________________________________ igt-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/igt-dev