- Linux-kselftest-mirror - lists.linaro.org

[PATCH v3 00/12] nolibc: add generic part1 of prepare for rv32

by Zhangjin Wu

Hi, Willy This is the v3 generic part1 for rv32, all of the found issues of v2 part1 [1] have been fixed up, several generic patches have been fixed up and merged from v2 part2 [2] to this series, the standalone test_fork patch [4] is merged with a Reviewed-by line into this series too. This series is based on 20230528-nolibc-rv32+stkp5 branch of [5]. Changes from v2 -> v3: * selftests/nolibc: fix up compile warning with glibc on x86_64 Use simpler 'long long' conversion instead of old #ifdef ... (Suggestion from Willy) * tools/nolibc: add missing nanoseconds support for __NR_statx Split the compound assignment into two single assignments (Suggestion from Thomas) * selftests/nolibc: add new gettimeofday test cases Removed the gettimeofday(NULL, &tz) (Suggestion from Thomas) All of the commit messages have been re-checked, some missing Suggested-by lines are added. The whole patchset have been tested on arm, aarch64, rv32 and rv64, no regressions (the next compile patchset is required to do rv32 test). The nolibc-test has been tested with glibc on x86_64 too. Btw, we have found such poll failures on arm (not introduced by this patchset), this will be fixed in our coming ppoll_time64 patchset: 48 poll_null = -1 ENOSYS [FAIL] 49 poll_stdout = -1 ENOSYS [FAIL] 50 poll_fault = -1 ENOSYS != (-1 EFAULT) [FAIL] And the gettimeofday_null removal patch from Thomas [3] may conflicts with the gettimeofday removal and addition patches, but it is not hard to fix. Best regards, Zhangjin --- [1]: https://lore.kernel.org/linux-riscv/cover.1685362482.git.falcon@tinylab.org… [2]: https://lore.kernel.org/linux-riscv/cover.1685387484.git.falcon@tinylab.org… [3]: https://lore.kernel.org/lkml/20230530-nolibc-gettimeofday-v1-1-7307441a002b… [4]: https://lore.kernel.org/lkml/61bdfe7bacebdef8aa9195f6f2550a5b0d33aab3.16854… [5]: https://git.kernel.org/pub/scm/linux/kernel/git/wtarreau/nolibc.git Zhangjin Wu (12): selftests/nolibc: syscall_args: use generic __NR_statx tools/nolibc: add missing nanoseconds support for __NR_statx selftests/nolibc: allow specify extra arguments for qemu selftests/nolibc: fix up compile warning with glibc on x86_64 selftests/nolibc: not include limits.h for nolibc selftests/nolibc: use INT_MAX instead of __INT_MAX__ tools/nolibc: arm: add missing my_syscall6 tools/nolibc: open: fix up compile warning for arm selftests/nolibc: support two errnos with EXPECT_SYSER2() selftests/nolibc: remove gettimeofday_bad1/2 completely selftests/nolibc: add new gettimeofday test cases selftests/nolibc: test_fork: fix up duplicated print tools/include/nolibc/arch-arm.h | 23 +++++++++++ tools/include/nolibc/stdint.h | 14 +++++++ tools/include/nolibc/sys.h | 39 +++++++++--------- tools/testing/selftests/nolibc/Makefile | 2 +- tools/testing/selftests/nolibc/nolibc-test.c | 42 ++++++++++++-------- 5 files changed, 85 insertions(+), 35 deletions(-) -- 2.25.1

2 years

3
18
0 0

[PATCH] tools/nolibc: ensure fast64 integer types have 64 bits

by Thomas Weißschuh

On 32bit platforms size_t is not enough to represent [u]int_fast64_t. Fixes: 3e9fd4e9a1d5 ("tools/nolibc: add integer types and integer limit macros") Signed-off-by: Thomas Weißschuh <linux(a)weissschuh.net> --- Cc: Vincent Dagonneau <v(a)vda.io> Note: We could also fall back to compiler-provided data like: __UINT_FAST{8,16,32,64}_{TYPE,MIN,MAX}__ --- tools/include/nolibc/stdint.h | 4 ++-- tools/testing/selftests/nolibc/nolibc-test.c | 6 +++--- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/tools/include/nolibc/stdint.h b/tools/include/nolibc/stdint.h index c1ce4f5e0603..3fc418cfc3d7 100644 --- a/tools/include/nolibc/stdint.h +++ b/tools/include/nolibc/stdint.h @@ -36,8 +36,8 @@ typedef ssize_t int_fast16_t; typedef size_t uint_fast16_t; typedef ssize_t int_fast32_t; typedef size_t uint_fast32_t; -typedef ssize_t int_fast64_t; -typedef size_t uint_fast64_t; +typedef int64_t int_fast64_t; +typedef uint64_t uint_fast64_t; typedef int64_t intmax_t; typedef uint64_t uintmax_t; diff --git a/tools/testing/selftests/nolibc/nolibc-test.c b/tools/testing/selftests/nolibc/nolibc-test.c index 7de46305f419..65be0317d184 100644 --- a/tools/testing/selftests/nolibc/nolibc-test.c +++ b/tools/testing/selftests/nolibc/nolibc-test.c @@ -696,9 +696,9 @@ int run_stdlib(int min, int max) CASE_TEST(limit_int_fast32_min); EXPECT_EQ(1, INT_FAST32_MIN, (int_fast32_t) INTPTR_MIN); break; CASE_TEST(limit_int_fast32_max); EXPECT_EQ(1, INT_FAST32_MAX, (int_fast32_t) INTPTR_MAX); break; CASE_TEST(limit_uint_fast32_max); EXPECT_EQ(1, UINT_FAST32_MAX, (uint_fast32_t) UINTPTR_MAX); break; - CASE_TEST(limit_int_fast64_min); EXPECT_EQ(1, INT_FAST64_MIN, (int_fast64_t) INTPTR_MIN); break; - CASE_TEST(limit_int_fast64_max); EXPECT_EQ(1, INT_FAST64_MAX, (int_fast64_t) INTPTR_MAX); break; - CASE_TEST(limit_uint_fast64_max); EXPECT_EQ(1, UINT_FAST64_MAX, (uint_fast64_t) UINTPTR_MAX); break; + CASE_TEST(limit_int_fast64_min); EXPECT_EQ(1, INT_FAST64_MIN, (int_fast64_t) INT64_MIN); break; + CASE_TEST(limit_int_fast64_max); EXPECT_EQ(1, INT_FAST64_MAX, (int_fast64_t) INT64_MAX); break; + CASE_TEST(limit_uint_fast64_max); EXPECT_EQ(1, UINT_FAST64_MAX, (uint_fast64_t) UINT64_MAX); break; #if __SIZEOF_LONG__ == 8 CASE_TEST(limit_intptr_min); EXPECT_EQ(1, INTPTR_MIN, (intptr_t) 0x8000000000000000LL); break; CASE_TEST(limit_intptr_max); EXPECT_EQ(1, INTPTR_MAX, (intptr_t) 0x7fffffffffffffffLL); break; --- base-commit: 5b21219d67d3483144d10332709d0c04f733ab93 change-id: 20230530-nolibc-fast64-8777cdbc7e01 Best regards, -- Thomas Weißschuh <linux(a)weissschuh.net>

2 years

3
3
0 0

[PATCH] selftests/nolibc: test_fork: fix up duplicated print

by Zhangjin Wu

running nolibc-test with glibc on x86_64 got such print issue: 29 execve_root = -1 EACCES [OK] 30 fork30 fork = 0 [OK] 31 getdents64_root = 712 [OK] The fork test case has three printf calls: (1) llen += printf("%d %s", test, #name); (2) llen += printf(" = %d %s ", expr, errorname(errno)); (3) llen += pad_spc(llen, 64, "[FAIL]\n"); --> vfprintf() In the following scene, the above issue happens: (a) The parent calls (1) (b) The parent calls fork() (c) The child runs and shares the print buffer of (1) (d) The child exits, flushs the print buffer and closes its own stdout/stderr * "30 fork" is printed at the first time. (e) The parent calls (2) and (3), with "\n" in (3), it flushs the whole buffer * "30 fork = 0 ..." is printed Therefore, there are two "30 fork" in the stdout. Between (a) and (b), if flush the stdout (and the sterr), the child in stage (c) will not be able to 'see' the print buffer. Signed-off-by: Zhangjin Wu <falcon(a)tinylab.org> --- tools/testing/selftests/nolibc/nolibc-test.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/tools/testing/selftests/nolibc/nolibc-test.c b/tools/testing/selftests/nolibc/nolibc-test.c index 7de46305f419..88323a60aa4a 100644 --- a/tools/testing/selftests/nolibc/nolibc-test.c +++ b/tools/testing/selftests/nolibc/nolibc-test.c @@ -486,7 +486,13 @@ static int test_getpagesize(void) static int test_fork(void) { int status; - pid_t pid = fork(); + pid_t pid; + + /* flush the printf buffer to avoid child flush it */ + fflush(stdout); + fflush(stderr); + + pid = fork(); switch (pid) { case -1: -- 2.25.1

2 years

3
3
0 0

[PATCH 0/1] selftests: error out if kernel header files are not yet built

by John Hubbard

Hi, This is somewhat related to the 11-patch series that I just posted [1], but I hadn't originally planned to go this far with it. But since David Hildenbrand asked if we could warn in this case [2], here it is. It turns out that automatically doing the "make headers" correctly is much harder than just warning, so I stopped at that point. It works well, though. [1] https://lore.kernel.org/all/20230603021558.95299-1-jhubbard@nvidia.com/ [2] https://lore.kernel.org/all/a4fbc191-9acb-5db8-a375-96c0c1ba3fcd@redhat.com/ thanks, John Hubbard Cc: David Hildenbrand <david(a)redhat.com> Cc: Peter Xu <peterx(a)redhat.com> Cc: Muhammad Usama Anjum <usama.anjum(a)collabora.com> Cc: Jonathan Corbet <corbet(a)lwn.net> Cc: linux-doc(a)vger.kernel.org John Hubbard (1): selftests: error out if kernel header files are not yet built tools/testing/selftests/lib.mk | 36 +++++++++++++++++++++++++++++++--- 1 file changed, 33 insertions(+), 3 deletions(-) base-commit: e5282a7d8f6b604f2bb6a06457734b8cf1e2f8f2 -- 2.40.1

2 years

1
1
0 0

[PATCH] Fix gitignore for recently added usptream self tests

by Weihao Gao

This resolves the issue that generated binary is showing up as an untracked git file after every build on the kernel. Signed-off-by: Weihao Gao <weihaogao(a)google.com> --- tools/testing/selftests/net/.gitignore | 3 +++ 1 file changed, 3 insertions(+) diff --git a/tools/testing/selftests/net/.gitignore b/tools/testing/selftests/net/.gitignore index 80f06aa62034..f27a7338b60e 100644 --- a/tools/testing/selftests/net/.gitignore +++ b/tools/testing/selftests/net/.gitignore @@ -8,8 +8,10 @@ diag_uid fin_ack_lat gro hwtstamp_config +io_uring_zerocopy_tx ioam6_parser ip_defrag +ip_local_port_range ipsec ipv6_flowlabel ipv6_flowlabel_mgr @@ -26,6 +28,7 @@ reuseport_bpf_cpu reuseport_bpf_numa reuseport_dualstack rxtimestamp +sctp_hello sk_bind_sendto_listen sk_connect_zero_addr socket -- 2.41.0.rc0.172.g3f132b7071-goog

2 years, 1 month

2
1
0 0

[PATCH] selftests/x86/amx: Add a CPU hotplug test

by Chang S. Bae

Adamos found the issue with the cached XFD state [1]. Although the XFD state is reset on the CPU hotplug, the per-CPU XFD cache is missing the reset. Then, running an AMX thread there, the staled value causes the kernel crash to kill the thread. This is reproducible when moving an AMX thread to the hot-plugged CPU. So, add a test case to ensure no issue with that. It repeats the test due to possible inconsistencies. Then, along with the hotplug cost, it will bring a noticeable runtime increase. But, the overall test has a quick turnaround time. Link: https://lore.kernel.org/lkml/20230519112315.30616-1-attofari@amazon.de/ [1] Signed-off-by: Chang S. Bae <chang.seok.bae(a)intel.com> Cc: Thomas Gleixner <tglx(a)linutronix.de> Cc: Dave Hansen <dave.hansen(a)linux.intel.com> Cc: Adamos Ttofari <attofari(a)amazon.de> Cc: Shuah Khan <shuah(a)kernel.org> Cc: linux-kselftest(a)vger.kernel.org Cc: linux-kernel(a)vger.kernel.org --- The overall x86 selftest via "$ make TARGETS='x86' kselftest" takes about 3.5 -> 5.5 seconds. 'amx_64' itself took about 1.5 more seconds over 0.x seconds. But, this overall runtime still takes in a matter of some seconds, which should be fine I thought. --- tools/testing/selftests/x86/amx.c | 133 ++++++++++++++++++++++++++++-- 1 file changed, 126 insertions(+), 7 deletions(-) diff --git a/tools/testing/selftests/x86/amx.c b/tools/testing/selftests/x86/amx.c index d884fd69dd51..6f2f0598c706 100644 --- a/tools/testing/selftests/x86/amx.c +++ b/tools/testing/selftests/x86/amx.c @@ -3,6 +3,7 @@ #define _GNU_SOURCE #include <err.h> #include <errno.h> +#include <fcntl.h> #include <pthread.h> #include <setjmp.h> #include <stdio.h> @@ -25,6 +26,8 @@ # error This test is 64-bit only #endif +#define BUF_LEN 1000 + #define XSAVE_HDR_OFFSET 512 #define XSAVE_HDR_SIZE 64 @@ -239,11 +242,10 @@ static inline uint64_t get_fpx_sw_bytes_features(void *buffer) } /* Work around printf() being unsafe in signals: */ -#define SIGNAL_BUF_LEN 1000 -char signal_message_buffer[SIGNAL_BUF_LEN]; +char signal_message_buffer[BUF_LEN]; void sig_print(char *msg) { - int left = SIGNAL_BUF_LEN - strlen(signal_message_buffer) - 1; + int left = BUF_LEN - strlen(signal_message_buffer) - 1; strncat(signal_message_buffer, msg, left); } @@ -767,15 +769,15 @@ static int create_threads(int num, struct futex_info *finfo) return 0; } -static void affinitize_cpu0(void) +static inline void affinitize_cpu(int cpu) { cpu_set_t cpuset; CPU_ZERO(&cpuset); - CPU_SET(0, &cpuset); + CPU_SET(cpu, &cpuset); if (sched_setaffinity(0, sizeof(cpuset), &cpuset) != 0) - fatal_error("sched_setaffinity to CPU 0"); + fatal_error("sched_setaffinity to CPU %d", cpu); } static void test_context_switch(void) @@ -784,7 +786,7 @@ static void test_context_switch(void) int i; /* Affinitize to one CPU to force context switches */ - affinitize_cpu0(); + affinitize_cpu(0); req_xtiledata_perm(); @@ -926,6 +928,121 @@ static void test_ptrace(void) err(1, "ptrace test"); } +/* CPU Hotplug test */ + +static void __hotplug_cpu(int online, int cpu) +{ + char buf[BUF_LEN] = {}; + int fd, rc; + + strncat(buf, "/sys/devices/system/cpu/cpu", BUF_LEN); + snprintf(buf + strlen(buf), BUF_LEN - strlen(buf), "%d", cpu); + strncat(buf, "/online", BUF_LEN - strlen(buf)); + + fd = open(buf, O_RDWR); + if (fd == -1) + fatal_error("open()"); + + snprintf(buf, BUF_LEN, "%d", online); + rc = write(fd, buf, strlen(buf)); + if (rc == -1) + fatal_error("write()"); + + rc = close(fd); + if (rc == -1) + fatal_error("close()"); +} + +static void offline_cpu(int cpu) +{ + __hotplug_cpu(0, cpu); +} + +static void online_cpu(int cpu) +{ + __hotplug_cpu(1, cpu); +} + +static jmp_buf jmpbuf; + +static void handle_sigsegv(int sig, siginfo_t *si, void *ctx_void) +{ + siglongjmp(jmpbuf, 1); +} + +#define RETRY 5 + +/* + * Sanity-check the hotplug CPU for its (re-)initialization. + * + * Create an AMX thread on a CPU, while the hotplug CPU went offline. + * Then, plug the offlined back, and move the thread to run on it. + * + * Repeat this multiple times to ensure no inconsistent failure. + * If something goes wrong, the thread will get a signal or killed. + */ +static void *switch_cpus(void *arg) +{ + int *result = (int *)arg; + int i = 0; + + affinitize_cpu(0); + offline_cpu(1); + load_rand_tiledata(stashed_xsave); + + sethandler(SIGSEGV, handle_sigsegv, SA_ONSTACK); + for (i = 0; i < RETRY; i++) { + if (i > 0) { + affinitize_cpu(0); + offline_cpu(1); + } + if (sigsetjmp(jmpbuf, 1) == 0) { + online_cpu(1); + affinitize_cpu(1); + } else { + *result = 1; + goto out; + } + } + *result = 0; +out: + clearhandler(SIGSEGV); + return result; +} + +static void test_cpuhp(void) +{ + int max_cpu_num = sysconf(_SC_NPROCESSORS_ONLN) - 1; + void *thread_retval; + pthread_t thread; + int result, rc; + + if (!max_cpu_num) { + printf("[SKIP]\tThe running system has no more CPU for the hotplug test.\n"); + return; + } + + printf("[RUN]\tTest AMX use with the CPU hotplug.\n"); + + if (pthread_create(&thread, NULL, switch_cpus, &result)) + fatal_error("pthread_create()"); + + rc = pthread_join(thread, &thread_retval); + + if (rc) + fatal_error("pthread_join()"); + + /* + * Either an invalid retval or a failed result indicates + * the test failure. + */ + if (thread_retval != &result || result != 0) + printf("[FAIL]\tThe AMX thread had an issue (%s).\n", + thread_retval != &result ? "killed" : "signaled"); + else + printf("[OK]\tThe AMX thread had no issue.\n"); +} + int main(void) { /* Check hardware availability at first */ @@ -948,6 +1065,8 @@ int main(void) test_ptrace(); + test_cpuhp(); + clearhandler(SIGILL); free_stashed_xsave(); base-commit: 7877cb91f1081754a1487c144d85dc0d2e2e7fc4 -- 2.17.1

2 years, 1 month

1
0
0 0

[PATCH 0/2] drivers: base: Add tests showing devm handling inconsistencies

by Maxime Ripard

Hi, This follows the discussion here: https://lore.kernel.org/linux-kselftest/20230324123157.bbwvfq4gsxnlnfwb@hou… This shows a couple of inconsistencies with regard to how device-managed resources are cleaned up. Basically, devm resources will only be cleaned up if the device is attached to a bus and bound to a driver. Failing any of these cases, a call to device_unregister will not end up in the devm resources being released. We had to work around it in DRM to provide helpers to create a device for kunit tests, but the current discussion around creating similar, generic, helpers for kunit resumed interest in fixing this. This can be tested using the command: ./tools/testing/kunit/kunit.py run --kunitconfig=drivers/base/test/ Let me know what you think, Maxime Signed-off-by: Maxime Ripard <maxime(a)cerno.tech> --- Maxime Ripard (2): drivers: base: Add basic devm tests for root devices drivers: base: Add basic devm tests for platform devices drivers/base/test/.kunitconfig | 2 + drivers/base/test/Kconfig | 4 + drivers/base/test/Makefile | 3 + drivers/base/test/platform-device-test.c | 278 +++++++++++++++++++++++++++++++ drivers/base/test/root-device-test.c | 120 +++++++++++++ 5 files changed, 407 insertions(+) --- base-commit: a6faf7ea9fcb7267d06116d4188947f26e00e57e change-id: 20230329-kunit-devm-inconsistencies-test-5e5a7d01e60d Best regards, -- Maxime Ripard <maxime(a)cerno.tech>

2 years, 1 month

3
8
0 0

Wyższa konwersja w e-sklepie

by Kamil Durjasz

Dzień dobry, w jaki sposób docierają Państwo do odbiorców? Tworzymy potężne narzędzia sprzedaży, które pozwalają kompleksowo rozwiązać problemy potencjalnych klientów i skutecznie wpłynąć na ich decyzje zakupowe. Skupiamy się na Państwa potrzebach związanych z obsługą sklepu, oczekiwaniach i planach sprzedażowych. Szczegółowo dopasowujemy grafikę, funkcjonalności, strukturę i mikrointerakcje do Państwa grupy docelowej, co przekłada się na oczekiwane rezultaty. Chętnie przedstawię dotychczasowe realizacje, aby mogli Państwo przekonać się o naszych możliwościach. Mogę się skontaktować? Pozdrawiam Kamil Durjasz

2 years, 1 month

1
0
0 0

[PATCH bpf v2 0/2] Fix elem_size not being set for inner maps

by Rhys Rustad-Elliott

Commit d937bc3449fa ("bpf: make uniform use of array->elem_size everywhere in arraymap.c") changed array_map_gen_lookup to use array->elem_size instead of round_up(map->value_size, 8) as the element size when generating code to access a value in an array map. array->elem_size, however, is not set by bpf_map_meta_alloc when initializing an BPF_MAP_TYPE_ARRAY_OF_MAPS or BPF_MAP_TYPE_HASH_OF_MAPS. This results in array_map_gen_lookup incorrectly outputting code that always accesses index 0 in the array (as the index will be calculated via a multiplication with the element size, which is incorrectly set to 0). This patchset sets elem_size on the bpf_array object when allocating an array or hash of maps to fix this and adds a selftest that accesses an array map nested within a hash of maps at a nonzero index to prevent regressions. v1: https://lore.kernel.org/bpf/95b5da7c-ee52-3ecb-0a4e-f6a7a114f269@linux.dev/ Changelog: v1 -> v2: Address comments by Martin KaFai Lau: - Directly use inner_array->elem_size instead of using round_up - Move selftests to a new patch - Use ASSERT_* macros instead of CHECK and remove duration - Remove unnecessary usleep - Shorten selftest name Rhys Rustad-Elliott (2): bpf: Fix elem_size not being set for inner maps selftests/bpf: Add access_inner_map selftest kernel/bpf/map_in_map.c | 8 +++- .../bpf/prog_tests/inner_array_lookup.c | 31 +++++++++++++ .../bpf/progs/test_inner_array_lookup.c | 45 +++++++++++++++++++ 3 files changed, 82 insertions(+), 2 deletions(-) create mode 100644 tools/testing/selftests/bpf/prog_tests/inner_array_lookup.c create mode 100644 tools/testing/selftests/bpf/progs/test_inner_array_lookup.c -- 2.40.1

2 years, 1 month

2
3
0 0

[PATCH] Documentation: kunit: Rename references to kunit_abort()

by David Gow

The kunit_abort() function has been renamed __kunit_abort(), update the references to it in the documentation. Suggested-by: Daniel Latypov <dlatypov(a)google.com> Signed-off-by: David Gow <davidgow(a)google.com> --- Documentation/dev-tools/kunit/architecture.rst | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Documentation/dev-tools/kunit/architecture.rst b/Documentation/dev-tools/kunit/architecture.rst index e95ab05342bb..f335f883f8f6 100644 --- a/Documentation/dev-tools/kunit/architecture.rst +++ b/Documentation/dev-tools/kunit/architecture.rst @@ -119,9 +119,9 @@ All expectations/assertions are formatted as: terminated immediately. - Assertions call the function: - ``void __noreturn kunit_abort(struct kunit *)``. + ``void __noreturn __kunit_abort(struct kunit *)``. - - ``kunit_abort`` calls the function: + - ``__kunit_abort`` calls the function: ``void __noreturn kunit_try_catch_throw(struct kunit_try_catch *try_catch)``. - ``kunit_try_catch_throw`` calls the function: -- 2.41.0.rc2.161.g9c6817b8e7-goog

2 years, 1 month

2
1
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-kselftest-mirror