On Wed, Apr 30, 2026 at 03:24:18PM +0200, Hector Cao wrote:
> Commit c551fb0b53d regressed module-loading behavior.
> Historically, when loading from one module directory failed,
> QEMU continued searching the remaining directories. That commit
> made QEMU stop if a matching file was found but could not be
> loaded (for example, due to a build mismatch).

Thanks for the v2, Hector. The errp handling with error_report_err()
looks correct now — errors are reported and cleared between iterations,
and the last error is propagated to the caller.

One minor edge case: if an earlier directory has a build mismatch
(rv = -1) but the retained module directory doesn't have the file
at all (ENOENT), rv stays at -1 with *errp == NULL. The caller
gets a failure return without an error message. In practice this
is unlikely since /run/qemu/<version>/ should contain the module
if it was retained, so I don't think this needs to block the patch.

Reviewed-by: Gabriel Hartmann <[email protected]>

-- 
Gabriel Hartmann
Senior Systems Engineer

NETWAYS Managed Services GmbH | Deutschherrnstr. 15-19 | D-90429 Nuernberg
Tel: +49 911 92885-0 | Fax: +49 911 92885-77
CEO: Julian Hein, Bernd Erk, Sebastian Saemann | AG Nuernberg HRB25207
https://www.netways.de | [email protected]

** Nuremberg Digital Festival | Open Source Night 23.06.26 | 
https://netways.de/nuedigital2026 **
** Open Tech Day 2026 | Software Defined Storage - October | Nuremberg - 
https://opentechday.de/ **
** Open Source Monitoring Conference 2026 - November | Nuremberg - 
https://osmc.de **
** NETWAYS Web Services - https://nws.netways.de **
** NETWAYS Trainings - https://netways.de/trainings **

Reply via email to