This is the mail archive of the glibc-bugs@sourceware.org mailing list for the glibc 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]

[Bug nptl/21016] pthread_cond support is broken on hppa


https://sourceware.org/bugzilla/show_bug.cgi?id=21016

--- Comment #3 from dave.anglin at bell dot net ---
On 2017-01-02, at 3:21 PM, triegel at redhat dot com wrote:

> Carlos has said he would take care of it.  I'm aware of two issues blocking
> this: atomics don't work in all cases (stores should use the kernel helper
> too), and the additional compatibility requirements for old initializers
> (IIRC).  To get the condvar files to compile, just copy over the new
> pthread_cond_t members from, say, x86 pthreadtypes.h to hppa's pthreadtypes.h.


I had a successful build by copying over the pthread_cond_t initial member from
x86.
I kept the old hppa member for the moment to get the same current alignment. 
Fortunately,
the sizes are the same (48).  I updated PTHREAD_COND_INITIALIZER to default.

I removed the LinuxThreads compatibility code although this is still a work in
progress:

        modified:   sysdeps/hppa/nptl/bits/pthreadtypes.h
        deleted:    sysdeps/unix/sysv/linux/hppa/internaltypes.h
        modified:   sysdeps/unix/sysv/linux/hppa/pthread.h
        deleted:    sysdeps/unix/sysv/linux/hppa/pthread_cond_broadcast.c
        deleted:    sysdeps/unix/sysv/linux/hppa/pthread_cond_destroy.c
        modified:   sysdeps/unix/sysv/linux/hppa/pthread_cond_init.c
        deleted:    sysdeps/unix/sysv/linux/hppa/pthread_cond_signal.c
        deleted:    sysdeps/unix/sysv/linux/hppa/pthread_cond_wait.c

Would like to get rid of pthread_cond_init.c and use default.

There is essentially zero code still using LinuxThreads in Debian, so I can't
see any reason to
try and retain compatibility with it.  With new union, I don't see how to
reliably check for the
LinuxThreads initializer.

In gcc, we use the kernel helper for stores so hopefully implementing this
shouldn't be hard.

--
John David Anglin       dave.anglin@bell.net

-- 
You are receiving this mail because:
You are on the CC list for the bug.

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