selftests/bpf: fix task_local_storage/exit_creds rcu usage
Subsystems (3)
PATCH v2 1/1 selftests/bpf: fix task_local_storage/exit_creds rcu usage
BPF CI has revealed flakiness in the task_local_storage/exit_creds test. The failure point in CI [1] is that null_ptr_count is equal to 0, which indicates that the program hasn't run yet. This points to the kern_sync_rcu (sys_membarrier -> synchronize_rcu underneath) not waiting sufficiently.
Indeed, synchronize_rcu only waits for read-side sections that started before the call. If the program execution starts *during* the synchronize_rcu invocation (due to, say, preemption), the test won't wait long enough.
As a speculative fix, make the synchornize_rcu calls in a loop until an explicit run counter has gone up.
[1]: https://github.com/kernel-patches/bpf/actions/runs/3268263235/jobs/5374940791
---
v1 -> v2:
Explicit loop counter and MAX_SYNC_RCU_CALLS guard.
.../bpf/prog_tests/task_local_storage.c | 18 +++++++++++++++--- .../bpf/progs/task_local_storage_exit_creds.c | 3 +++ 2 files changed, 18 insertions(+), 3 deletions(-)
tools/testing/selftests/bpf/prog_tests/task_local_storage.c +15 -3
@@ -39,7 +39,8 @@ static void test_sys_enter_exit(void)static void test_exit_creds (void ) {struct task_local_storage_exit_creds *skel; -int err; +int err, run_count, sync_rcu_calls =0 ; +const int MAX_SYNC_RCU_CALLS =1000 ; skel =task_local_storage_exit_creds__open_and_load ();if (!ASSERT_OK_PTR (skel,"skel_open_and_load" ))
@@ -53,8 +54,19 @@ static void test_exit_creds(void)if (CHECK_FAIL (system ("ls > /dev/null" ))) goto out; -/* sync rcu to make sure exit_creds() is called for "ls" */ -kern_sync_rcu (); +/* kern_sync_rcu is not enough on its own as the read section we want +* to wait for may start after we enter synchronize_rcu, so our call +* won't wait for the section to finish. Loop on the run counter +* as well to ensure the program has run. +*/ +do { +kern_sync_rcu (); + run_count =__atomic_load_n (&skel->bss->run_count, __ATOMIC_SEQ_CST); + }while (run_count ==0 && ++sync_rcu_calls <MAX_SYNC_RCU_CALLS ); + +ASSERT_NEQ (sync_rcu_calls,MAX_SYNC_RCU_CALLS , +"sync_rcu count too high" ); +ASSERT_NEQ (run_count,0 ,"run_count" );ASSERT_EQ (skel->bss->valid_ptr_count,0 ,"valid_ptr_count" );ASSERT_NEQ (skel->bss->null_ptr_count,0 ,"null_ptr_count" ); out:
tools/testing/selftests/bpf/progs/task_local_storage_exit_creds.c +3 -0
@@ -14,6 +14,7 @@ struct {__type (value, __u64); } task_storage SEC(".maps"); +int run_count =0 ;int valid_ptr_count =0 ;int null_ptr_count =0 ;
@@ -28,5 +29,7 @@ int BPF_PROG(trace_exit_creds, struct task_struct *task)__sync_fetch_and_add (&valid_ptr_count,1 ); else __sync_fetch_and_add(&null_ptr_count, 1); + +__sync_fetch_and_add (&run_count,1 );return 0 ; }
Signature · 1 line
2.37.3
Quoted 45 lines · Delyan Kratunov
BPF CI has revealed flakiness in the task_local_storage/exit_creds test. The failure point in CI [1] is that
null_ptr_countis equal to 0, which indicates that the program hasn't run yet. This points to thekern_sync_rcu(sys_membarrier->synchronize_rcuunderneath) not waiting sufficiently.Indeed,
synchronize_rcuonly waits for read-side sections that started before the call. If the program execution starts *during* thesynchronize_rcuinvocation (due to, say, preemption), the test won't wait long enough.As a speculative fix, make the
synchornize_rcucalls in a loop until an explicit run counter has gone up.[1]:https://github.com/kernel-patches/bpf/actions/runs/3268263235/jobs/5374940791
Signed-off-by Delyan Kratunov <[email protected]>---
v1 -> v2:
Explicit loop counter andMAX_SYNC_RCU_CALLSguard..../bpf/prog_tests/task_local_storage.c | 18 +++++++++++++++--- .../bpf/progs/task_local_storage_exit_creds.c | 3 +++ 2 files changed, 18 insertions(+), 3 deletions(-)tools/testing/selftests/bpf/prog_tests/task_local_storage.c +2
-2@@ -39,7 +39,8 @@ static void test_sys_enter_exit(void)static void test_exit_creds (void ) {struct task_local_storage_exit_creds *skel; -int err; +int err, run_count, sync_rcu_calls =0 ; +const int MAX_SYNC_RCU_CALLS =1000 ; skel =task_local_storage_exit_creds__open_and_load ();if (!ASSERT_OK_PTR (skel,"skel_open_and_load" ))@@ -53,8 +54,19 @@ static void test_exit_creds(void)if (CHECK_FAIL (system ("ls > /dev/null" ))) goto out; -/* sync rcu to make sure exit_creds() is called for "ls" */
tools/testing/selftests/bpf/prog_tests/task_local_storage.c static void test_exit_creds(void) · line 65
Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/task_local_storage.c +9
-1-kern_sync_rcu (); +/* kern_sync_rcu is not enough on its own as the read section we want +* to wait for may start after we enter synchronize_rcu, so our call +* won't wait for the section to finish. Loop on the run counter +* as well to ensure the program has run. +*/ +do { +kern_sync_rcu (); + run_count =__atomic_load_n (&skel->bss->run_count, __ATOMIC_SEQ_CST); + }while (run_count ==0 && ++sync_rcu_calls <MAX_SYNC_RCU_CALLS );
Might have been easier to do the following instead?
int sync_rcu_calls = 1000;
do {
} while (run_count == 0 && --sync_rcu_calls);
Quoted 29 lines
tools/testing/selftests/bpf/prog_tests/task_local_storage.c +4
-0+ +ASSERT_NEQ (sync_rcu_calls,MAX_SYNC_RCU_CALLS , +"sync_rcu count too high" ); +ASSERT_NEQ (run_count,0 ,"run_count" );ASSERT_EQ (skel->bss->valid_ptr_count,0 ,"valid_ptr_count" );ASSERT_NEQ (skel->bss->null_ptr_count,0 ,"null_ptr_count" ); out:tools/testing/selftests/bpf/progs/task_local_storage_exit_creds.c +3
-1@@ -14,6 +14,7 @@ struct {__type (value, __u64); } task_storage SEC(".maps"); +int run_count =0 ;int valid_ptr_count =0 ;int null_ptr_count =0 ;@@ -28,5 +29,7 @@ int BPF_PROG(trace_exit_creds, struct task_struct *task)__sync_fetch_and_add (&valid_ptr_count,1 ); else __sync_fetch_and_add(&null_ptr_count, 1); + +__sync_fetch_and_add (&run_count,1 );return 0 ; } --2.37.3
Quoted 59 lines · On Wed, Oct 19, 2022 at 12:57 PM <[email protected]> wrote:
Quoted 46 lines · Delyan Kratunov
BPF CI has revealed flakiness in the task_local_storage/exit_creds test. The failure point in CI [1] is that
null_ptr_countis equal to 0, which indicates that the program hasn't run yet. This points to thekern_sync_rcu(sys_membarrier->synchronize_rcuunderneath) not waiting sufficiently.Indeed,
synchronize_rcuonly waits for read-side sections that started before the call. If the program execution starts *during* thesynchronize_rcuinvocation (due to, say, preemption), the test won't wait long enough.As a speculative fix, make the
synchornize_rcucalls in a loop until an explicit run counter has gone up.[1]:https://github.com/kernel-patches/bpf/actions/runs/3268263235/jobs/5374940791
Signed-off-by Delyan Kratunov <[email protected]>---
v1 -> v2:
Explicit loop counter andMAX_SYNC_RCU_CALLSguard..../bpf/prog_tests/task_local_storage.c | 18 +++++++++++++++--- .../bpf/progs/task_local_storage_exit_creds.c | 3 +++ 2 files changed, 18 insertions(+), 3 deletions(-)tools/testing/selftests/bpf/prog_tests/task_local_storage.c +2
-2@@ -39,7 +39,8 @@ static void test_sys_enter_exit(void)static void test_exit_creds (void ) {struct task_local_storage_exit_creds *skel; -int err; +int err, run_count, sync_rcu_calls =0 ; +const int MAX_SYNC_RCU_CALLS =1000 ; skel =task_local_storage_exit_creds__open_and_load ();if (!ASSERT_OK_PTR (skel,"skel_open_and_load" ))@@ -53,8 +54,19 @@ static void test_exit_creds(void)if (CHECK_FAIL (system ("ls > /dev/null" ))) goto out; -/* sync rcu to make sure exit_creds() is called for "ls" */ tools/testing/selftests/bpf/prog_tests/task_local_storage.c static void test_exit_creds(void) · line 65
Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/task_local_storage.c +9
-1-kern_sync_rcu (); +/* kern_sync_rcu is not enough on its own as the read section we want +* to wait for may start after we enter synchronize_rcu, so our call +* won't wait for the section to finish. Loop on the run counter +* as well to ensure the program has run. +*/ +do { +kern_sync_rcu (); + run_count =__atomic_load_n (&skel->bss->run_count, __ATOMIC_SEQ_CST); + }while (run_count ==0 && ++sync_rcu_calls <MAX_SYNC_RCU_CALLS );
Quoted 7 lines
Might have been easier to do the following instead?
int
sync_rcu_calls= 1000;
do {
} while (run_count== 0 && --sync_rcu_calls);
I think it's a preference of the author.
Both are fine. imo.
I was about to apply, but then noticed Delyan's author line and SOB are different. @meta vs @fb :(
Delyan, please fix.
Fixing SOB is not something maintainers can do while applying.
Quoted 65 lines · Alexei Starovoitov
Quoted 63 lines · On Wed, Oct 19, 2022 at 12:57 PM <[email protected]> wrote:
Quoted 46 lines · Delyan Kratunov
BPF CI has revealed flakiness in the task_local_storage/exit_creds test. The failure point in CI [1] is that
null_ptr_countis equal to 0, which indicates that the program hasn't run yet. This points to thekern_sync_rcu(sys_membarrier->synchronize_rcuunderneath) not waiting sufficiently.Indeed,
synchronize_rcuonly waits for read-side sections that started before the call. If the program execution starts *during* thesynchronize_rcuinvocation (due to, say, preemption), the test won't wait long enough.As a speculative fix, make the
synchornize_rcucalls in a loop until an explicit run counter has gone up.[1]:https://github.com/kernel-patches/bpf/actions/runs/3268263235/jobs/5374940791
Signed-off-by Delyan Kratunov <[email protected]>---
v1 -> v2:
Explicit loop counter andMAX_SYNC_RCU_CALLSguard..../bpf/prog_tests/task_local_storage.c | 18 +++++++++++++++--- .../bpf/progs/task_local_storage_exit_creds.c | 3 +++ 2 files changed, 18 insertions(+), 3 deletions(-)tools/testing/selftests/bpf/prog_tests/task_local_storage.c +2
-2@@ -39,7 +39,8 @@ static void test_sys_enter_exit(void)static void test_exit_creds (void ) {struct task_local_storage_exit_creds *skel; -int err; +int err, run_count, sync_rcu_calls =0 ; +const int MAX_SYNC_RCU_CALLS =1000 ; skel =task_local_storage_exit_creds__open_and_load ();if (!ASSERT_OK_PTR (skel,"skel_open_and_load" ))@@ -53,8 +54,19 @@ static void test_exit_creds(void)if (CHECK_FAIL (system ("ls > /dev/null" ))) goto out; -/* sync rcu to make sure exit_creds() is called for "ls" */ tools/testing/selftests/bpf/prog_tests/task_local_storage.c static void test_exit_creds(void) · line 65
Quoted 10 lines
tools/testing/selftests/bpf/prog_tests/task_local_storage.c +9
-1-kern_sync_rcu (); +/* kern_sync_rcu is not enough on its own as the read section we want +* to wait for may start after we enter synchronize_rcu, so our call +* won't wait for the section to finish. Loop on the run counter +* as well to ensure the program has run. +*/ +do { +kern_sync_rcu (); + run_count =__atomic_load_n (&skel->bss->run_count, __ATOMIC_SEQ_CST); + }while (run_count ==0 && ++sync_rcu_calls <MAX_SYNC_RCU_CALLS );Might have been easier to do the following instead?
Quoted 7 lines
Quoted 3 lines
int
sync_rcu_calls= 1000;
do {
} while (run_count== 0 && --sync_rcu_calls);I think it's a preference of the author.
Both are fine. imo.
Agreed, that's why I acked it, shouldn't really matter.
Quoted 5 lines
I was about to apply, but then noticed Delyan's author line and SOB are different. @meta vs @fb :(
Delyan, please fix.Fixing SOB is not something maintainers can do while applying.