NetBSD-Bugs archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
re: kern/56245: NetBSD 9.2 amd64 with AMD FirePro W4100 throws constant kernel errors of 'Can't enable IRQ/MSI because no handler is installed'
The following reply was made to PR kern/56245; it has been noted by GNATS.
From: matthew green <mrg%eterna.com.au@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: kern-bug-people%netbsd.org@localhost, gnats-admin%netbsd.org@localhost,
netbsd-bugs%netbsd.org@localhost
Subject: re: kern/56245: NetBSD 9.2 amd64 with AMD FirePro W4100 throws constant kernel errors of 'Can't enable IRQ/MSI because no handler is installed'
Date: Mon, 14 Jun 2021 08:11:09 +1000
this is strange.
> Jun 4 18:33:45 vorlon /netbsd: [ 8.6896180] radeon0: info: radeon: MS=
I limited to 32-bit
> Jun 4 18:33:45 vorlon /netbsd: [ 8.6896180] radeon0: interrupting at =
ioapic0 pin 16 (radeon0)
> Jun 4 18:33:45 vorlon /netbsd: [ 8.6896180] kern info: [drm] radeon: =
irq initialized.
so, MSI is not enabled, otherwise this message wuld appear:
dev_info(rdev->dev, "radeon: using MSI.\n");
> Jun 4 18:33:45 vorlon /netbsd: [ 9.8597595] warning: /usr/src/sys/ext=
ernal/bsd/drm2/dist/drm/radeon/radeon_si.c:6101: Can't enable IRQ/MSI beca=
use no handler is installed
so, this seems impossible. that message happens if
rdev->irq.installed is not set in si_irq_set().
si_startup() checks if it's not enabled, and then
calls radeon_irq_kms_init() to set it, which sets
installed =3D true, so the call a few lines later to
si_irq_set() should have it set, instead of the
error message it displays and doesn't setup.
i don't see a path that doesn't do this, but i guess
there's something i'm missing.
> spamming syslog messages.
> [ 428896.996373] kern error: [drm:(/usr/src/sys/external/bsd/drm2/dist/d=
rm/radeon/radeon_si_dpm.c:6697)si_dpm_set_power_state] *ERROR* si_enable_s=
mc_cac fail ed
this happens on r600 as well, i never got around to look at why.
.mrg.
Home |
Main Index |
Thread Index |
Old Index