[Bug nptl/19986] sparc32 glibc build fails with "No support for pthread barriers on pre-v9 sparc"

["triegel at redhat dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=19986

--- Comment #3 from Torvald Riegel <triegel at redhat dot com> ---
pre-v9 sparc doesn't have full support for atomic operations (ie, no
compare-and-set, just test-and-set).  Therefore, concurrent code often has to
be adapted.  Other concurrent data structures have custom versions, but it
could be possible to do this in a way that allows for code reuse with the
generic version used for all targets.  It would also be possible to change the
implementation of the sparc pre-v9 atomics, so that it falls back on a generic
lock.  These are the points the sparc maintainer should decide, IMO.

You could look at other concurrent data structures to see the differences
between sparc pre-v9 and generic code, if you want to give it a try.  You could
also just disable barriers if you don't need them for your use case.

If you have a stronger interest in pre-v9 sparc, and have some cycles, it could
also be worthwhile if you could help with sparc maintenance of glibc.  The
current sparc maintainer doesn't seem to have a lot of cycles to spend.

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