gdb.base/info-os.exp FAILs/ERRORs [Re: [commit] 'info os' additions]
Stan Shebs
stanshebs@earthlink.net
Sat May 12 20:18:00 GMT 2012
On 5/12/12 8:40 AM, Jan Kratochvil wrote:
> On Sat, 12 May 2012 03:32:12 +0200, Stan Shebs wrote:
>> * gdb.base/info-os.exp: New file.
>> * gdb.base/info-os.c: New file.
> With heavy parallelization it gives:
>
> +ERROR: internal buffer is full.
> +UNRESOLVED: gdb.base/info-os.exp: get process list
> +ERROR: internal buffer is full.
> +UNRESOLVED: gdb.base/info-os.exp: get process groups
> +ERROR: internal buffer is full.
> +UNRESOLVED: gdb.base/info-os.exp: get threads
> +ERROR: internal buffer is full.
> +UNRESOLVED: gdb.base/info-os.exp: get file descriptors
> +ERROR: internal buffer is full.
> +UNRESOLVED: gdb.base/info-os.exp: get internet-domain sockets
>
> where the buffer overflows get fixed by the attached patch. But even with that
> patch it still produces:
>
> info os semaphores^M
> (gdb) FAIL: gdb.base/info-os.exp: get semaphores
> info os msg^M
> (gdb) FAIL: gdb.base/info-os.exp: get message queues
> (gdb) info os shm^M
> key shmid permissions size creator command last op. command num attached user group creator user creator group last shmat() time last shmdt() time last shmctl() time ^M
> 3925 196608 666 4096 info-os 0 0 jkratoch jkratoch jkratoch jkratoch Sat May 12 17:22:10 2012 ^M
> (gdb) FAIL: gdb.base/info-os.exp: get shared-memory regions
>
> as apparently the .c testfile creates constant resources and expects they are
> free, therefore multiple runs clash. There can be even multiple runs by
> different users. This is why for example gdbserver_start does:
> # Loop till we find a free port.
> [...]
> incr portnum
>
> Could you fix this part?
Huh, didn't think about parallelization! I'll fix it up.
Stan
>
>
> Thanks,
> Jan
>
> gdb/testsuite/
> 2012-05-12 Jan Kratochvil<jan.kratochvil@redhat.com>
>
> Fix ERROR: internal buffer is full.
> * gdb.base/info-os.exp (expect_multiline): New function.
> (get process list, get process groups, get threads)
> (get file descriptors, get internet-domain sockets)
> (get shared-memory regions, get semaphores, get message queues): Use
> it for these tests.
>
> diff --git a/gdb/testsuite/gdb.base/info-os.exp b/gdb/testsuite/gdb.base/info-os.exp
> index e1f7224..5696402 100644
> --- a/gdb/testsuite/gdb.base/info-os.exp
> +++ b/gdb/testsuite/gdb.base/info-os.exp
> @@ -90,15 +90,62 @@ gdb_test_multiple "print port" $test {
> }
> }
>
> +# Act like gdb_test but prevent: +ERROR: internal buffer is full.
> +
> +proc expect_multiline { command expect test } {
> + global gdb_prompt
> +
> + # Do not duplicate FAILs from gdb_test_multiple.
> + set pass 0
> + set fail 0
> + gdb_test_multiple $command $test {
> + -re "^$expect *\r\n" {
> + pass $test
> + set pass 1
> + exp_continue
> + }
> + -re "^$gdb_prompt $" {
> + if !$pass {
> + set fail 1
> + }
> + # Exit the loop.
> + }
> + -re "\r\n" {
> + # Drop the buffer.
> + exp_continue
> + }
> + }
> + if $fail {
> + fail $test
> + }
> +}
> +
> # Test output of the 'info os' commands against the expected results.
> -gdb_test "info os processes" ".*pid +user +command +cores.*$inferior_pid +\\S+ +\\S*info-os +\[0-9\]+.*" "get process list"
> -gdb_test "info os procgroups" ".*pgid +leader command +pid +command line.*$inferior_pid +info-os +$inferior_pid +\\S*info-os.*" "get process groups"
> -gdb_test "info os threads" ".*pid +command +tid +core.*$inferior_pid +info-os +\\d+ +\\d+.*" "get threads"
> -gdb_test "info os files" ".*pid +command +file descriptor +name.*$inferior_pid +info-os +\\d+ +/dev/null.*" "get file descriptors"
> -gdb_test "info os sockets" ".*local address +local port +remote address +remote port +state +user +family +protocol.*0\\.0\\.0\\.0 +$port +0\\.0\\.0\\.0 +0 +LISTEN +\\S+ +INET +STREAM.*" "get internet-domain sockets"
> -gdb_test "info os shm" ".*key +shmid +permissions +size +creator command +last op\\. command +num attached +user +group +creator user +creator group +last shmat\\(\\) time +last shmdt\\(\\) time +last shmctl\\(\\) time.*3925 +$shmid +666 +4096 +info-os +.*" "get shared-memory regions"
> -gdb_test "info os semaphores" ".*key +semid +permissions +num semaphores +user +group +creator user +creator group +last semop\\(\\) time +last semctl\\(\\) time.*7428 +$semid +666 +1 +.*" "get semaphores"
> -gdb_test "info os msg" ".*key +msqid +permissions +num used bytes +num messages +last msgsnd\\(\\) command +last msgrcv\\(\\) command +user +group +creator user +creator group +last msgsnd\\(\\) time +last msgrcv\\(\\) time +last msgctl\\(\\) time.*5294 +$msqid +666 +.*" "get message queues"
> +
> +# pid user command cores
> +expect_multiline "info os processes" "$inferior_pid +\\S+ +\\S*info-os +\[0-9,\]+" "get process list"
> +
> +# pgid leader pid command line
> +expect_multiline "info os procgroups" "$inferior_pid +info-os +$inferior_pid +\\S*info-os" "get process groups"
> +
> +# pid command tid core
> +expect_multiline "info os threads" "$inferior_pid +info-os +\\d+ +\\d+" "get threads"
> +
> +# pid command fd name
> +expect_multiline "info os files" "$inferior_pid +info-os +\\d+ +/dev/null" "get file descriptors"
> +
> +# local address l-port remote addr r-port state user family protocol
> +expect_multiline "info os sockets" "0\\.0\\.0\\.0 +$port +0\\.0\\.0\\.0 +0 +LISTEN +\\S+ +INET +STREAM" "get internet-domain sockets"
> +
> +# key shmid perm size creator command last op command num attached user group creator user creator group last shmat() time last shmdt() time last shmctl() time
> +expect_multiline "info os shm" "3925 +$shmid +666 +4096 +info-os .*" "get shared-memory regions"
> +
> +# key semid perm num semaphores user group creator user creator group last semop() time last semctl() time
> +expect_multiline "info os semaphores" "7428 +$semid +666 +1 .*" "get semaphores"
> +
> +# key msqid perm num used bytes num messages last msgsnd() command last msgrcv() command user group creator user creator group last msgsnd() time last msgrcv() time last msgctl() time
> +expect_multiline "info os msg" "5294 +$msqid +666 .*" "get message queues"
> +
>
> # The SysV IPC primitives linger on after the creating process is killed
> # unless they are destroyed explicitly, so allow the test program to tidy
>
More information about the Gdb-patches
mailing list