Hi,
On 2025-12-04 12:36, Graham Inggs wrote:
> Source: openvswitch
> Version: 3.6.0-5
> Severity: serious
> Tags: ftbfs
> X-Debbugs-Cc: [email protected]
> User: [email protected]
> Usertags: riscv64
>
> Hi Maintainer
>
> The build of openvswitch frequently fails on riscv64 [1], but seems
> fine on other architectures.
>
> Please investigate if this can be improved.
I have been able to reproduce the issue as part of the build, where many
tests are run in parallel, but not running this specific bfd.at test in
a loop. It seems to be quite sensitive to the timing, and it also fails
from time to time, although with a lower frequency, on s390x.
It seems that the failing part of the test is already disabled by
upstream on arm64, so as a first step I suggest to do the same on
riscv64. That would be this patch:
--- openvswitch-3.6.0.orig/tests/atlocal.in
+++ openvswitch-3.6.0/tests/atlocal.in
@@ -104,6 +104,9 @@ case `uname -m` in
aarch64)
IS_ARM64="yes"
;;
+riscv64)
+ IS_RISCV64="yes"
+ ;;
*)
IS_ARM64="no"
;;
--- openvswitch-3.6.0.orig/tests/bfd.at
+++ openvswitch-3.6.0/tests/bfd.at
@@ -267,6 +267,7 @@ AT_CLEANUP
# Tests below are for bfd decay features.
AT_SETUP([bfd - bfd decay])
AT_SKIP_IF([test "$IS_ARM64" = "yes"])
+AT_SKIP_IF([test "$IS_RISCV64" = "yes"])
OVS_VSWITCHD_START([add-br br1 -- set bridge br1 datapath-type=dummy -- \
add-port br1 p1 -- set Interface p1 type=patch \
options:peer=p0 ofport_request=2 -- \
Regards
Aurelien
--
Aurelien Jarno GPG: 4096R/1DDD8C9B
[email protected] http://aurel32.net
