On 12/29/2016 07:20 AM, Khem Raj wrote:
On Tue, Dec 27, 2016 at 11:46 PM, Chen Qi <[email protected]> wrote:
rsyslog uses a small program to check if compiler has atomic builtins
or not. This does not work when cross compiling, resulting in potential
race issue. In fact, when running ptest on intel-x86-64 board, we have
met the race problem.

As our compiler gcc, has atomic builtins, we force ap_cv_atomic_builtins
to be 'yes' so that rsyslog could make use of the atomic builtins and avoid
the race issue.

Signed-off-by: Chen Qi <[email protected]>
---
  meta-oe/recipes-extended/rsyslog/rsyslog_8.22.0.bb | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/meta-oe/recipes-extended/rsyslog/rsyslog_8.22.0.bb 
b/meta-oe/recipes-extended/rsyslog/rsyslog_8.22.0.bb
index 4e2c122..27162e1 100644
--- a/meta-oe/recipes-extended/rsyslog/rsyslog_8.22.0.bb
+++ b/meta-oe/recipes-extended/rsyslog/rsyslog_8.22.0.bb
@@ -31,7 +31,7 @@ SRC_URI[sha256sum] = 
"06e2884181333dccecceaca82827ae24ca7a258b4fbf7b1e07a80d4caa

  inherit autotools pkgconfig systemd update-rc.d update-alternatives ptest

-EXTRA_OECONF += "--disable-generate-man-pages"
+EXTRA_OECONF += "--disable-generate-man-pages ap_cv_atomic_builtins=yes"
Does this also cover 64bit builtins ? secondly, a note for
architectures not supporting atomics to disable this would be good.

No. When looking at the source, I think it's not critical.
When I made this patch, my criteria is:
1) ptest of rsyslog does not hang because of race issue on intel-x86-64 board 2) rsyslog works well on qemux86, qemux86-64, qemuarm, qemumips, qemumips64, qemuppc

I searched the doc and the internet, I could not find related information on which arch gcc does not support atomic builtins.

Best Regards,
Chen Qi

  # first line is default yes in configure
  PACKAGECONFIG ??= " \
--
2.8.3

--
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel


--
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel

Reply via email to