Hi Guinevere,
We've found and fixed an issue in Linaro CI.
The problem was that when a regression is detected (solib-precsave.exp failures in this case), the report also included pre-existing known failures in the regression list. So your 1/8 patch indeed has an issue that needs fixing, but only solib-precsave.exp failures are actual regressions.
Thanks for bringing this issue to our attention!
Kind regards,
-- Maxim Kuvyrkov https://www.linaro.org
On Jun 9, 2026, at 14:39, Guinevere Larsen guinevere@redhat.com wrote:
On 6/9/26 12:52 AM, Thiago Jung Bauermann wrote:
Thiago Jung Bauermann thiago.bauermann@linaro.org writes:
The FAILs in gdb.threads/ are almost surely regressions not yet marked as such by our CI. I also see the gdb.base/gdb-sigterm.exp and gdb.mi/mi-break.exp FAILs on current trunk so they're also false positives.
Sorry, I mischecked mi-break.exp. It passes fully on current trunk. It's common for MI tests to be flaky though, so it could also be a case of a flaky test not yet flagged as such by our CI.
Yeah, based on what is changed by the test, there's nothing outside of gdb.reverse that would be affected by this test
the solib-precsave.exp failures are definitely concerning though, thanks for pointing those out!
-- Cheers, Guinevere Larsen It/she