This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Recent 4.1 breakage...


Howard Hinnant wrote:

However, the semaphore is not lost forever, it may be manually removed
with ipcrm(1). Might be a FreeBSD QoI issue. OTOH, since semaphores
span processes, it is probably working as intended. I don't know offhand.

Frankly, that seems to me a QoI issue: seems very annoying that if random processes using semaphores start failing for some reason, then sempahore are not available anymore for the *whole* system...

Since:

2006-03-12 Howard Hinnant <hhinnant@apple.com>

   * testsuite/27_io/basic_filebuf/underflow/char/10097.cc:
   Fix race condition.
   * testsuite/27_io/objects/char/9661-1.cc: Likewise.
   * testsuite/27_io/objects/wchar_t/9661-1.cc: Likewise.

increased the use of semaphores (by 1 in each of these tests), that explains why this change may have caused this type of failure to show up at a lower N test count.

That is not possible, because the patch is only in mainline.


Anyway, remains to explain why 26777 is failing in the first place on FreeBSD. Without hints I have no idea, sorry.

Paolo.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]