lib/igt_core: Add igt_system helpers
PATCH v3 1/3 lib/igt_core: Add igt_system helpers
Support executing external processes with the goal of capturing its standard streams to the igt logging infrastructure in addition to its exit status.
v3: Rename igt_exec -> igt_system (Chris).
v2: Fix leaks on fd teardown. Make sure redirected process printout when
Quoted 1 line
64kb still works, like full dmesg. (Petri).
--- lib/igt_core.c | 157 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ lib/igt_core.h | 10 ++++ 2 files changed, 167 insertions(+)
lib/igt_core.c +157 -0
@@ -2099,3 +2099,160 @@ FILE *__igt_fopen_data(const char* igt_srcdir, const char* igt_datadir,return fp; } + +struct output_pipe { +int output_fd; +int save_fd; +int read_fd; +int write_fd; +bool redirected; +enum igt_log_level log_level; +}; + +static bool redirect_output (struct output_pipe *p,int output_fd, +enum igt_log_level level) +{ +int fds[2 ]; + +if (pipe (fds) ==-1 ) + goto err; + +/* save output */ +if ((p->save_fd =dup (output_fd)) ==-1 ) + goto err; + +/* Redirect output to our buffer */ +if (dup2 (fds[1 ], output_fd) ==-1 ) + goto err; + + p->output_fd = output_fd; + p->read_fd = fds[0 ]; + p->write_fd = fds[1 ]; + p->redirected = true; + p->log_level = level; + +return true; +err: +close (fds[0 ]); +close (fds[1 ]); +close (p->save_fd); + +return false; +} + +static void unredirect_output (struct output_pipe *p) +{ +if (!p->redirected) +return ; + +/* read_fd is closed separately. We still need to read its +* buffered contents after un-redirecting the stream. +*/ +close (p->write_fd); +dup2 (p->save_fd, p->output_fd); +close (p->save_fd); + p->redirected = false; +} + +/** +* igt_system: +* +* An improved replacement of the system() call. +* +* Executes the shell command specified in @command with the added feature of +* concurrently capturing its stdout and stderr to igt_log and igt_warn +* respectively. +* +* Returns: The exit status of the executed process. -1 for failure. +*/ +int igt_system (const char *command) +{ +#define OUT 0 +#define ERR 1 +struct output_pipe op[2 ]; +int i, status; +struct igt_helper_process process = {}; +char buf[PIPE_BUF ]; + +if (!redirect_output (&op[OUT ],STDOUT_FILENO ,IGT_LOG_INFO )) + goto err; +if (!redirect_output (&op[ERR ],STDERR_FILENO ,IGT_LOG_WARN )) + goto err; + + igt_fork_helper(&process) { +igt_assert (execl ("/bin/sh" ,"sh" ,"-c" , command, + (char *)NULL ) !=-1 ); + } + +for (i =0 ; i <ARRAY_SIZE (op); i++) { +struct output_pipe *current = &op[i]; + +/* Unredirect so igt_log() works */ +unredirect_output (current); +memset (buf,0 ,sizeof (buf)); +while (read (current->read_fd, buf,sizeof (buf)) >0 ) { +igt_log (IGT_LOG_DOMAIN , current->log_level, +"[cmd] %s" , buf); +memset (buf,0 ,sizeof (buf)); + } +close (current->read_fd); + } + status =igt_wait_helper (&process); + +return WEXITSTATUS (status); +err: +/* Failed to redirect one or both streams. Roll back changes. */ +for (i =0 ; i <ARRAY_SIZE (op); i++) { +if (!op[i].redirected) +continue ; +close (op[i].read_fd); +unredirect_output (&op[i]); + } + +return -1 ; +} + +/** +* igt_system_quiet: +* Similar to igt_system(), except redirect output to /dev/null +* +* Returns: The exit status of the executed process. -1 for failure. +*/ +int igt_system_quiet (const char *command) +{ +int stderr_fd_copy, stdout_fd_copy, status, nullfd; + +/* redirect */ +if ((nullfd =open ("/dev/null" ,O_WRONLY )) ==-1 ) + goto err; +if ((stdout_fd_copy =dup (STDOUT_FILENO )) ==-1 ) + goto err; +if ((stderr_fd_copy =dup (STDERR_FILENO )) ==-1 ) + goto err; + +if (dup2 (nullfd,STDOUT_FILENO ) ==-1 ) + goto err; +if (dup2 (nullfd,STDERR_FILENO ) ==-1 ) + goto err; + +if ((status =system (command)) ==-1 ) + goto err; + +/* restore */ +if (dup2 (stdout_fd_copy,STDOUT_FILENO ) ==-1 ) + goto err; +if (dup2 (stderr_fd_copy,STDERR_FILENO ) ==-1 ) + goto err; + +close (stdout_fd_copy); +close (stderr_fd_copy); +close (nullfd); + +return WEXITSTATUS (status); +err: +close (stderr_fd_copy); +close (stdout_fd_copy); +close (nullfd); + +return -1 ; +}
lib/igt_core.h +10 -0
@@ -919,4 +919,14 @@ FILE *__igt_fopen_data(const char* igt_srcdir, const char* igt_datadir,#define igt_fopen_data (filename) \ __igt_fopen_data(IGT_SRCDIR, IGT_DATADIR, filename) +int igt_system (const char *command); +int igt_system_quiet (const char *command); +#define igt_system_cmd(status, format...) \ + do { \ + char *buf = 0; \ + igt_assert(asprintf(&buf, format) != -1); \ + status = igt_system(buf); \ + free(buf); \ + } while (0) +#endif /* IGT_CORE_H */
Signature · 6 lines
2.7.4 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/3 igt/igt_core: Provide an option to check for the log buffer contents
--- lib/igt_core.c | 24 ++++++++++++++++++++++++ lib/igt_core.h | 3 +++ 2 files changed, 27 insertions(+)
lib/igt_core.c +24 -0
@@ -329,6 +329,30 @@ static void _igt_log_buffer_dump(void)pthread_mutex_unlock (&log_buffer_mutex); } +/** +* igt_log_buffer_inspect: +* +* Provides a way to replay the internal igt log buffer for inspection. +* @check: A user-specified handler that gets invoked for each line of +the log buffer. The handler should return true to stop +inspecting the rest of the buffer. +* @data: passed as a user argument to the inspection function. +*/ +void igt_log_buffer_inspect (igt_buffer_log_handler_t check,void *data) +{ +uint8_t i; +pthread_mutex_lock (&log_buffer_mutex); + + i = log_buffer.start; +do { +if (check (log_buffer.entries[i], data)) +break ; + i++; + }while (i != log_buffer.start && i != log_buffer.end); + +pthread_mutex_unlock (&log_buffer_mutex); +} + __attribute__((format (printf,1 ,2 ))) static void kmsg(const char *format, ...)#define KERN_EMER "<0>"
lib/igt_core.h +3 -0
@@ -818,6 +818,9 @@ void igt_vlog(const char *domain, enum igt_log_level level, const char *format,*/ #define igt_critical(f...) igt_log(IGT_LOG_DOMAIN, IGT_LOG_CRITICAL, f) +typedef bool (*igt_buffer_log_handler_t )(const char *line,void *data); +void igt_log_buffer_inspect (igt_buffer_log_handler_t check,void *data); +extern enum igt_log_level igt_log_level;/**
Signature · 6 lines
2.7.4 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [i-g-t PATCH 2/3] igt/igt_core: Provide an option to check for the log buffer contents
Quoted 11 lines · Abdiel Janulgue
Signed-off-by Abdiel Janulgue <[email protected]>--- lib/igt_core.c | 24 ++++++++++++++++++++++++ lib/igt_core.h | 3 +++ 2 files changed, 27 insertions(+)lib/igt_core.c +0
-0
lib/igt_core.c static void _igt_log_buffer_dump(void) · line 338
Quoted 10 lines
lib/igt_core.c +7
-0@@ -329,6 +329,30 @@ static void _igt_log_buffer_dump(void)pthread_mutex_unlock (&log_buffer_mutex); } +/** +* igt_log_buffer_inspect: +* +* Provides a way to replay the internal igt log buffer for inspection. +* @check: A user-specified handler that gets invoked for each line of +the log buffer. The handler should return true to stop +inspecting the rest of the buffer.
Looks like we have an asterisk even for the indented lines elsewhere.
With the above nitpick
Signature · 6 lines
Cheers, Arek _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 3/3 Convert shell script tests to C version
In reply to Abdiel Janulguev3: Drop redundant test covered by drv_hangman/basic. Descend thru
debugfs path when reading sysfs entries (Chris).
v2: Use internal igt_debugfs functions instead of cat and document
debugfs tests.
Convert sysfs_l3_parity properly.
Rename redundant names in tests.
Converted:
- check_drm_clients (ensures no other clients are running. functionality provided by drm_open_driver_master).
- debugfs_emon_crash
- debugfs_wedged
- drv_debugfs_reader
- sysfs_l3_parity
- test_rte_check (same as check_drm_clients)
- tools_test
- ZZ_check_dmesg
--- tests/Makefile.sources | 9 +--- tests/ZZ_check_dmesg | 11 ----- tests/check_drm_clients | 6 --- tests/debugfs.c | 96 +++++++++++++++++++++++++++++++++++++ tests/debugfs_emon_crash | 16 ------- tests/debugfs_wedged | 10 ---- tests/drv_debugfs_reader | 9 ---- tests/sysfs_l3_parity | 22 --------- tests/test_rte_check | 6 --- tests/tools.c | 122 +++++++++++++++++++++++++++++++++++++++++++++++ tests/tools_test | 16 ------- 11 files changed, 220 insertions(+), 103 deletions(-) delete mode 100755 tests/ZZ_check_dmesg delete mode 100755 tests/check_drm_clients create mode 100644 tests/debugfs.c delete mode 100755 tests/debugfs_emon_crash delete mode 100755 tests/debugfs_wedged delete mode 100755 tests/drv_debugfs_reader delete mode 100755 tests/sysfs_l3_parity delete mode 100755 tests/test_rte_check create mode 100644 tests/tools.c delete mode 100755 tests/tools_test
tests/Makefile.sources +2 -7
@@ -243,6 +243,8 @@ TESTS_progs = \ drv_module_reload \ kms_sysfs_edid_timing \ perf \ + debugfs \ + tools \ $(NULL) # IMPORTANT: The ZZ_ tests need to be run last!
@@ -251,11 +253,6 @@ TESTS_scripts_M = \ $(NULL) TESTS_scripts = \ - debugfs_emon_crash \ - drv_debugfs_reader \ - sysfs_l3_parity \ - test_rte_check \ - tools_test \ $(NULL) # This target contains testcases which support automagic subtest enumeration
@@ -317,9 +314,7 @@ HANG = \ $(NULL) scripts = \ - check_drm_clients \ ddx_intel_after_fbdev \ - debugfs_wedged \ drm_lib.sh \ drm_getopt.sh \ $(NULL)
tests/ZZ_check_dmesg deleted +0 -11
@@ -1,11 +0,0 @@ -#!/bin/sh - -if dmesg | grep '\*ERROR\*' > /dev/null ; then - echo "DRM_ERROR dirt in dmesg" - exit 1 -fi - -if dmesg | grep -- '------\[ cut here \]----' > /dev/null ; then - echo "found a backtrace in dmesg" - exit 1 -fi
tests/check_drm_clients deleted +0 -6
@@ -1,6 +0,0 @@ -#!/bin/bash - -SOURCE_DIR="$( dirname "${BASH_SOURCE[0]}" )" -. $SOURCE_DIR/drm_lib.sh - -exit $IGT_EXIT_SUCCESS
tests/debugfs.c new file +96 -0
96 lines not on this page · raw message
tests/debugfs_emon_crash deleted +0 -16
@@ -1,16 +0,0 @@ -#!/bin/bash -# -# This check if we can crash the kernel with segmentation-fault -# by reading /sys/kernel/debug/dri/0/i915_emon_status too quickly -# - -SOURCE_DIR="$( dirname "${BASH_SOURCE[0]}" )" -. $SOURCE_DIR/drm_lib.sh - -for z in $(seq 1 1000); do - cat $i915_dfs_path/i915_emon_status > /dev/null 2&>1 -done - -# If we got here, we haven't crashed - -exit $IGT_EXIT_SUCCESS
tests/debugfs_wedged deleted +0 -10
@@ -1,10 +0,0 @@ -#!/bin/bash - -SOURCE_DIR="$( dirname "${BASH_SOURCE[0]}" )" -. $SOURCE_DIR/drm_lib.sh - -# Testcase: wedge the hw to check the error_state reading -# -# Unfortunately wedged is permanent, so this test is not run by default -echo 1 > ${i915_dfs_path}/i915_wedged -cat $i915_dfs_path/i915_error_state > /dev/null 2>&1
tests/drv_debugfs_reader deleted +0 -9
@@ -1,9 +0,0 @@ -#!/bin/bash - -SOURCE_DIR="$( dirname "${BASH_SOURCE[0]}" )" -. $SOURCE_DIR/drm_lib.sh - -# read everything we can -cat $i915_dfs_path/* > /dev/null 2>&1 - -exit $IGT_EXIT_SUCCESS
tests/sysfs_l3_parity deleted +0 -22
@@ -1,22 +0,0 @@ -#!/bin/bash - -SOURCE_DIR="$( dirname "${BASH_SOURCE[0]}" )" -. $SOURCE_DIR/drm_lib.sh - -$SOURCE_DIR/../tools/intel_l3_parity -r 0 -b 0 -s 0 -e || exit $? - -#Check that we can remap a row -$SOURCE_DIR/../tools/intel_l3_parity -r 0 -b 0 -s 0 -d -disabled=`$SOURCE_DIR/../tools/intel_l3_parity -l | grep -c 'Row 0, Bank 0, Subbank 0 is disabled'` -if [ "$disabled" != "1" ] ; then - echo "Fail" - exit $IGT_EXIT_FAILURE -fi - -$SOURCE_DIR/../tools/intel_l3_parity -r 0 -b 0 -s 0 -e - -#Check that we can clear remaps -if [ `$SOURCE_DIR/../tools/intel_l3_parity -l | wc -l` != 1 ] ; then - echo "Fail 2" - exit $IGT_EXIT_FAILURE -fi
tests/test_rte_check deleted +0 -6
@@ -1,6 +0,0 @@ -#!/bin/bash - -SOURCE_DIR="$( dirname "${BASH_SOURCE[0]}" )" -. $SOURCE_DIR/drm_lib.sh - -exit $IGT_EXIT_SUCCESS
tests/tools.c new file +122 -0
122 lines not on this page · raw message
tests/tools_test deleted +0 -16
@@ -1,16 +0,0 @@ -#!/bin/bash -# Test some of the most critical tools we have accidentally broken before. -# TODO: Possibly make tests parse output - -SOURCE_DIR="$( dirname "${BASH_SOURCE[0]}" )" -. $SOURCE_DIR/drm_lib.sh - -# ARB_MODE has existed for many gens -PATH=$SOURCE_DIR/../tools:$PATH -do_or_die "intel_reg read 0x4030" -do_or_die "intel_reg dump" - -# TODO: Add more tests - -exit $IGT_EXIT_SUCCESS -
Signature · 6 lines
2.7.4 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [i-g-t PATCH v3 3/3] Convert shell script tests to C version
Quoted 8 lines · Abdiel Janulgue
v3: Drop redundant test covered by drv_hangman/basic. Descend thru
debugfs path when reading sysfs entries (Chris).v2: Use internal
igt_debugfsfunctions instead of cat and document
debugfs tests.
Convertsysfs_l3_parityproperly.
Rename redundant names in tests.
Quoted 10 lines
Converted:
-check_drm_clients(ensures no other clients are running. functionality provided bydrm_open_driver_master).
-debugfs_emon_crash
-debugfs_wedged
-drv_debugfs_reader
-sysfs_l3_parity
-test_rte_check(same ascheck_drm_clients)
-tools_test
-ZZ_check_dmesg
Hey,
Doing each tool in a separate commit would help with readability.
Quoted 32 lines
Cc Petri Latvala <[email protected]>Signed-off-by Abdiel Janulgue <[email protected]>--- tests/Makefile.sources | 9 +--- tests/ZZ_check_dmesg | 11 ----- tests/check_drm_clients | 6 --- tests/debugfs.c | 96 +++++++++++++++++++++++++++++++++++++ tests/debugfs_emon_crash | 16 ------- tests/debugfs_wedged | 10 ---- tests/drv_debugfs_reader | 9 ---- tests/sysfs_l3_parity | 22 --------- tests/test_rte_check | 6 --- tests/tools.c | 122 +++++++++++++++++++++++++++++++++++++++++++++++ tests/tools_test | 16 ------- 11 files changed, 220 insertions(+), 103 deletions(-) delete mode 100755 tests/ZZ_check_dmesg delete mode 100755 tests/check_drm_clients create mode 100644 tests/debugfs.c delete mode 100755 tests/debugfs_emon_crash delete mode 100755 tests/debugfs_wedged delete mode 100755 tests/drv_debugfs_reader delete mode 100755 tests/sysfs_l3_parity delete mode 100755 tests/test_rte_check create mode 100644 tests/tools.c delete mode 100755 tests/tools_testtests/Makefile.sources +0
-0
tests/Makefile.sources TESTS_progs = \ · line 246
Quoted 4 lines
tests/Makefile.sources +1
-0@@ -243,6 +243,8 @@ TESTS_progs = \ drv_module_reload \ kms_sysfs_edid_timing \ perf \ + debugfs \
The name clashes on AOSP build:build/core/base_rules.mk:238: error: external/igt/tests: MODULE.TARGET.EXECUTABLES.debugfs already defined by external/e2fsprogs/debugfs.
debugfs is a name of a binary that debugs ext{2,3,4} family of FSes.
I have to look into the build system and our Android.mks more deeply as IGTs output binaries should be stored elsewhere and the clash is rather enigmatic.
Also are we changing names of the test? I thought it was only a rewrite.
Quoted 5 lines
tests/Makefile.sources +1
-0+ tools \ $(NULL) # IMPORTANT: The ZZ_ tests need to be run last!
tests/Makefile.sources TESTS_scripts_M = \ · line 256
Quoted 9 lines
tests/Makefile.sources +0
-5@@ -251,11 +253,6 @@ TESTS_scripts_M = \ $(NULL) TESTS_scripts = \ - debugfs_emon_crash \ - drv_debugfs_reader \ - sysfs_l3_parity \ - test_rte_check \ - tools_test \ $(NULL)
TESTS_scripts now is only $(NULL), so you may as well just remove it from the kernel_tests variable and get rid of it completely.
Quoted 12 lines
tests/Makefile.sources +0
-2# This target contains testcases which support automagic subtest enumeration@@ -317,9 +314,7 @@ HANG = \ $(NULL) scripts = \ - check_drm_clients \ ddx_intel_after_fbdev \ - debugfs_wedged \ drm_lib.sh \ drm_getopt.sh \ $(NULL)
<SNIP>
Quoted 28 lines
tests/debugfs.c new file +22
-0@@ -0,0 +1,96 @@ +/* +* Copyright © 2017 Intel Corporation +* +* Permission is hereby granted, free of charge, to any person obtaining a +* copy of this software and associated documentation files (the "Software"), +* to deal in the Software without restriction, including without limitation +* the rights to use, copy, modify, merge, publish, distribute, sublicense, +* and/or sell copies of the Software, and to permit persons to whom the +* Software is furnished to do so, subject to the following conditions: +* +* The above copyright notice and this permission notice (including the next +* paragraph) shall be included in all copies or substantial portions of the +* Software. +* +* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING +* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS +* IN THE SOFTWARE. +*/
tests/debugfs.c line 32
Quoted 10 lines
tests/debugfs.c new file +10
-0+#ifdef HAVE_CONFIG_H +#include "config.h" +#endif +#include "igt.h" +#include "igt_sysfs.h" +#include <fcntl.h> +#include <sys/types.h> +#include <dirent.h> + +static void get_sysfs_entry (int path_fd)
I was a little bit confused by the function - we are not getting anything from it. Just reading and discarding.
Maybe read_and_discard_sysfs_entry? A comment would also do.
Quoted 11 lines
tests/debugfs.c new file +11
-0+{ +struct dirent *dirent; +DIR *dir; + + dir =fdopendir (path_fd); +if (!dir) +return ; + +while ((dirent =readdir (dir))) { +if (!strcmp (dirent->d_name,"." ) || + !strcmp (dirent->d_name,".." ))
tests/debugfs.c line 53
Quoted 10 lines
tests/debugfs.c new file +10
-0+continue ; +if (dirent->d_type ==DT_DIR ) { +int sub_fd =-1 ; +igt_assert ((sub_fd = +openat (path_fd, dirent->d_name,O_RDONLY | +O_DIRECTORY )) >0 ); +get_sysfs_entry (sub_fd); +close (sub_fd); + }else { +char *buf =igt_sysfs_get (path_fd, dirent->d_name);
igt_sysfs_get may get us NULL.
Shouldn't we assert on that? It's an error-worthy.
Also we may end up trying to free it in with the free below.
Quoted 43 lines
tests/debugfs.c new file +43
-0+free (buf); + } + } +closedir (dir); +} + +igt_main +{ +int fd =-1 , debugfs; +igt_skip_on_simulation (); + + igt_fixture { + fd =drm_open_driver_master (DRIVER_INTEL ); +igt_require_gem (fd); + debugfs =igt_debugfs_dir (fd); + } + + igt_subtest("drv_reader") { +get_sysfs_entry (debugfs); + } + + igt_subtest("emon_crash") { +int i; +/* +* This check if we can crash the kernel with +* segmentation-fault by reading +* /sys/kernel/debug/dri/0/i915_emon_status too quickly +*/ +for (i =0 ; i <1000 ; i++) { +char *buf =igt_sysfs_get (debugfs, +"i915_emon_status" ); +free (buf); + } + +/* If we got here, we haven't crashed */ +igt_success (); + } + + igt_fixture { +close (debugfs); +close (fd); + } +}
<SNIP>
Quoted 116 lines
tests/tools.c new file +110
-0@@ -0,0 +1,122 @@ +/* +* Copyright © 2017 Intel Corporation +* +* Permission is hereby granted, free of charge, to any person obtaining a +* copy of this software and associated documentation files (the "Software"), +* to deal in the Software without restriction, including without limitation +* the rights to use, copy, modify, merge, publish, distribute, sublicense, +* and/or sell copies of the Software, and to permit persons to whom the +* Software is furnished to do so, subject to the following conditions: +* +* The above copyright notice and this permission notice (including the next +* paragraph) shall be included in all copies or substantial portions of the +* Software. +* +* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING +* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS +* IN THE SOFTWARE. +*/ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif +#include "igt.h" +#include <stdio.h> + +/** +* Parse the r-value of a [cmd] string. +*/ +static bool check_cmd_return_value (const char *s,void *data) +{ +int *val = data; +char *cmd, *found; +const char *delim ="[cmd]" ; +const int delim_len =strlen (delim); + +if (!(cmd =strstr (s, delim))) +return false; + + found = cmd + delim_len +1 ; +igt_assert (delim_len +strlen (found) <strlen (cmd)); + + *val =atoi (found); +return true; +} + +igt_main +{ +int fd =-1 ; +igt_skip_on_simulation (); + + igt_subtest("sysfs_l3_parity") { +int exec_return; + +igt_system_cmd (exec_return, +"../tools/intel_l3_parity -r 0 -b 0 " +"-s 0 -e" ); +igt_assert (exec_return ==IGT_EXIT_SUCCESS ); + +igt_system_cmd (exec_return, +"../tools/intel_l3_parity -l | " +"grep -c 'Row 0, Bank 0, Subbank 0 " +"is disabled'" ); +if (exec_return ==IGT_EXIT_SUCCESS ) { +int val =-1 ; +igt_log_buffer_inspect (check_cmd_return_value, + &val); +igt_assert (val ==1 ); + }else { +igt_fail (IGT_EXIT_FAILURE ); + } + +igt_system_cmd (exec_return, +"../tools/intel_l3_parity -r 0 -b 0 " +"-s 0 -e" ); +igt_assert (exec_return ==IGT_EXIT_SUCCESS ); + +/* Check that we can clear remaps */ +igt_system_cmd (exec_return, +"../tools/intel_l3_parity -l | " +"wc -l" ); +if (exec_return ==IGT_EXIT_SUCCESS ) { +int val =-1 ; +igt_log_buffer_inspect (check_cmd_return_value, + &val); +igt_assert (val ==1 ); + }else { +igt_fail (IGT_EXIT_FAILURE ); + } + } + + igt_subtest("tools_test") { +char *cmd; + +igt_assert (asprintf (&cmd, +"../tools/intel_reg read 0x4030" ) + !=-1 ); +igt_assert (igt_system_quiet (cmd) ==IGT_EXIT_SUCCESS ); +free (cmd); + +igt_assert (asprintf (&cmd,"../tools/intel_reg dump" ) + !=-1 ); +igt_assert (igt_system_quiet (cmd) ==IGT_EXIT_SUCCESS ); +free (cmd); + } + + igt_subtest("check_dmesg") { +char *cmd;
tests/tools.c line 120
Quoted 10 lines
tests/tools.c new file +10
-0+igt_assert (asprintf (&cmd,"dmesg | grep '\\*ERROR\\*'" ) + !=-1 ); +igt_assert (igt_system_quiet (cmd) !=IGT_EXIT_SUCCESS ); +free (cmd); + +igt_assert (asprintf (&cmd,"dmesg | grep " +"-- '------\\[ cut here \\]----'" ) + !=-1 ); +igt_assert (igt_system_quiet (cmd) !=IGT_EXIT_SUCCESS ); +free (cmd);
Are we testing the dmesg here?
If not they why systeming out "dmesg | grep" instead of going through /dev/kmsg?
Quoted 2 lines
tests/tools.c new file +2
-0+ } +}
Signature · 6 lines
Cheers, Arek _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [i-g-t PATCH v3 3/3] Convert shell script tests to C version
Quoted 20 lines · Arkadiusz Hiler
Quoted 8 lines · Abdiel Janulgue
v3: Drop redundant test covered by drv_hangman/basic. Descend thru
debugfs path when reading sysfs entries (Chris).v2: Use internal
igt_debugfsfunctions instead of cat and document
debugfs tests.
Convertsysfs_l3_parityproperly.
Rename redundant names in tests.Quoted 10 lines
Converted:
-check_drm_clients(ensures no other clients are running. functionality provided bydrm_open_driver_master).
-debugfs_emon_crash
-debugfs_wedged
-drv_debugfs_reader
-sysfs_l3_parity
-test_rte_check(same ascheck_drm_clients)
-tools_test
-ZZ_check_dmesg
Quoted 3 lines
Hey,
Doing each tool in a separate commit would help with readability.
Will do.
Quoted 41 lines
Quoted 33 lines
Cc Petri Latvala <[email protected]>Signed-off-by Abdiel Janulgue <[email protected]>--- tests/Makefile.sources | 9 +--- tests/ZZ_check_dmesg | 11 ----- tests/check_drm_clients | 6 --- tests/debugfs.c | 96 +++++++++++++++++++++++++++++++++++++ tests/debugfs_emon_crash | 16 ------- tests/debugfs_wedged | 10 ---- tests/drv_debugfs_reader | 9 ---- tests/sysfs_l3_parity | 22 --------- tests/test_rte_check | 6 --- tests/tools.c | 122 +++++++++++++++++++++++++++++++++++++++++++++++ tests/tools_test | 16 ------- 11 files changed, 220 insertions(+), 103 deletions(-) delete mode 100755 tests/ZZ_check_dmesg delete mode 100755 tests/check_drm_clients create mode 100644 tests/debugfs.c delete mode 100755 tests/debugfs_emon_crash delete mode 100755 tests/debugfs_wedged delete mode 100755 tests/drv_debugfs_reader delete mode 100755 tests/sysfs_l3_parity delete mode 100755 tests/test_rte_check create mode 100644 tests/tools.c delete mode 100755 tests/tools_testtests/Makefile.sources +0
-0tests/Makefile.sources TESTS_progs = \ · line 246
Quoted 4 lines
tests/Makefile.sources +1
-0@@ -243,6 +243,8 @@ TESTS_progs = \ drv_module_reload \ kms_sysfs_edid_timing \ perf \ + debugfs \The name clashes on AOSP build:
build/core/base_rules.mk:238: error: external/igt/tests: MODULE.TARGET.EXECUTABLES.debugfs already defined by external/e2fsprogs/debugfs.
Quoted 7 lines
debugfs is a name of a binary that debugs ext{2,3,4} family of FSes.
I have to look into the build system and our Android.mks more deeply as IGTs output binaries should be stored elsewhere and the clash is rather enigmatic.
Also are we changing names of the test? I thought it was only a rewrite.
It's just a rewrite. Now that we combine those debugfs_* scripts into one executable, maybe just the prefix would suffice? How about debugfs_tests.c?
Quoted 76 lines
Quoted 6 lines
tests/Makefile.sources +1
-0+ tools \ $(NULL) # IMPORTANT: The ZZ_ tests need to be run last!tests/Makefile.sources TESTS_scripts_M = \ · line 256
Quoted 9 lines
tests/Makefile.sources +0
-5@@ -251,11 +253,6 @@ TESTS_scripts_M = \ $(NULL) TESTS_scripts = \ - debugfs_emon_crash \ - drv_debugfs_reader \ - sysfs_l3_parity \ - test_rte_check \ - tools_test \ $(NULL)
TESTS_scriptsnow is only $(NULL), so you may as well just remove it from thekernel_testsvariable and get rid of it completely.Quoted 12 lines
tests/Makefile.sources +0
-2# This target contains testcases which support automagic subtest enumeration@@ -317,9 +314,7 @@ HANG = \ $(NULL) scripts = \ - check_drm_clients \ ddx_intel_after_fbdev \ - debugfs_wedged \ drm_lib.sh \ drm_getopt.sh \ $(NULL)<SNIP>
Quoted 28 lines
tests/debugfs.c new file +22
-0@@ -0,0 +1,96 @@ +/* +* Copyright © 2017 Intel Corporation +* +* Permission is hereby granted, free of charge, to any person obtaining a +* copy of this software and associated documentation files (the "Software"), +* to deal in the Software without restriction, including without limitation +* the rights to use, copy, modify, merge, publish, distribute, sublicense, +* and/or sell copies of the Software, and to permit persons to whom the +* Software is furnished to do so, subject to the following conditions: +* +* The above copyright notice and this permission notice (including the next +* paragraph) shall be included in all copies or substantial portions of the +* Software. +* +* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING +* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS +* IN THE SOFTWARE. +*/ tests/debugfs.c line 32
Quoted 10 lines
tests/debugfs.c new file +10
-0+#ifdef HAVE_CONFIG_H +#include "config.h" +#endif +#include "igt.h" +#include "igt_sysfs.h" +#include <fcntl.h> +#include <sys/types.h> +#include <dirent.h> + +static void get_sysfs_entry (int path_fd)I was a little bit confused by the function - we are not getting anything from it. Just reading and discarding.
Quoted 1 line
Maybe
read_and_discard_sysfs_entry? A comment would also do.
read_and_discard_sysfs_entry() sounds better.
Quoted 17 lines
Quoted 17 lines
tests/debugfs.c new file +16
-0+{ + struct dirent *dirent; + DIR *dir; + + dir = fdopendir(path_fd); + if (!dir) + return; + + while ((dirent = readdir(dir))) { + if (!strcmp(dirent->d_name, ".") || + !strcmp(dirent->d_name, "..")) + continue; + if (dirent->d_type == DT_DIR) { + int sub_fd = -1; + igt_assert((sub_fd = + openat(path_fd, dirent->d_name, O_RDONLY |
Quoted 8 lines
tests/debugfs.c line 53
Quoted 5 lines
tests/debugfs.c new file +5
-0+ O_DIRECTORY)) > 0); +get_sysfs_entry (sub_fd); +close (sub_fd); + }else { +char *buf =igt_sysfs_get (path_fd, dirent->d_name);
igt_sysfs_getmay get us NULL.
Shouldn't we assert on that? It's an error-worthy.
Yep.
Quoted 10 lines
<SNIP>
tests/debugfs.c
Quoted 5 lines
tests/debugfs.c new file +5
-0+igt_assert (asprintf (&cmd,"dmesg | grep " +"-- '------\\[ cut here \\]----'" ) + !=-1 ); +igt_assert (igt_system_quiet (cmd) !=IGT_EXIT_SUCCESS ); +free (cmd);Are we testing the dmesg here?
Quoted 2 lines
If not they why systeming out "dmesg | grep" instead of going through
/dev/kmsg?
Yes. This could be improved. Thanks for the review!
_______________________________________________
Intel-gfx mailing list [email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 11 lines · Abdiel Janulgue
Support executing external processes with the goal of capturing its standard streams to the igt logging infrastructure in addition to its exit status.
v3: Rename
igt_exec->igt_system(Chris).v2: Fix leaks on fd teardown. Make sure redirected process printout when
Quoted 1 line
64kb still works, like full dmesg. (Petri).
Cc Petri Latvala <[email protected]>Signed-off-by Abdiel Janulgue <[email protected]>
Signature · 6 lines
Cheers, Arek _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx