DMARC-Filter: OpenDMARC Filter v1.4.2 delorie.com 55JJEdV21790811
DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org DE1983857BB9
Date: Thu, 19 Jun 2025 12:13:22 -0700 (PDT)
To: cygwin@cygwin.com
Subject: Re: consistent stress test failures on arm64 after recent signal
 changes
In-Reply-To: <c3652c83-00a5-948c-ece1-687bcc38c8b7@jdrake.com>
Message-ID: <a365e5f4-4d15-a7c1-eebe-32edeeaa0e4d@jdrake.com>
References: <c3652c83-00a5-948c-ece1-687bcc38c8b7@jdrake.com>
MIME-Version: 1.0
Precedence: list
From: Jeremy Drake via Cygwin <cygwin@cygwin.com>
Reply-To: Jeremy Drake <cygwin@jdrake.com>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Errors-To: cygwin-bounces~archive-cygwin=delorie.com@cygwin.com
Sender: "Cygwin" <cygwin-bounces~archive-cygwin=delorie.com@cygwin.com>

On Wed, 18 Jun 2025, Jeremy Drake via Cygwin wrote:

> The "stress tests x86_64 on arm64 Windows" tests "fp" and "memcpy" seem to
> be consitently failing since 364226a.  The step is allowed to continue on
> error because there is some flakiness with tests on arm64 Windows, but
> these tests seem to be pretty consistently failing.
>
> It seems like 3 commits were pushed at once, so there's only one job for:
>
> Cygwin: signal: Prevent unexpected crash on frequent SIGSEGV
> Cygwin: signal: Revert _cygtls::inside_kernel() change
> Cygwin: signal: Enable the signal mask earlier
>
> See for example
> https://github.com/cygwin/cygwin/actions/runs/15706999164/job/44256840429#step:6:37

I tried reverting each of these, and found that the commit
f305ca916ad25870fb010334e4fcaf93057c78b9 ("Cygwin: signal: Prevent
unexpected crash on frequent SIGSEGV") is the one that causes these
failures.

-- 
Problem reports:      https://cygwin.com/problems.html
FAQ:                  https://cygwin.com/faq/
Documentation:        https://cygwin.com/docs.html
Unsubscribe info:     https://cygwin.com/ml/#unsubscribe-simple