Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 14 Nov 2020 08:39:42 -0500
From:      Alexander Motin <mav@FreeBSD.org>
To:        martin@vx.sk
Cc:        src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-12@freebsd.org
Subject:   Re: svn commit: r367359 - stable/12/sys/dev/pci
Message-ID:  <e0dd58bc-251b-b80c-b3b6-e4ebbbc00229@FreeBSD.org>
In-Reply-To: <20201113173821.Horde.jq1A9wp72tHUvM10uhYxhOq@mail.vx.sk>
References:  <20201113173821.Horde.jq1A9wp72tHUvM10uhYxhOq@mail.vx.sk>

next in thread | previous in thread | raw e-mail | index | archive | help
Martin,

I've never seen such issues.  Could you send me verbose `dmesg`,
`devinfo -v` and `lspci -vv` outputs with and without the tunable set?
I guess your BIOS sets some things differently from all I've tested.

On 13.11.2020 11:38, martin@vx.sk wrote:
> I am encountering a problem with hw.pci.enable_aspm:
> 
> System: ASUS RS500A-E10-RS12U
> CPU: AMD EPYC 7502 32-Core.
> NVMe drive model: Micron 9300 PRO
> Number of NVMe drives: 10
> 
> Using X2APIC
> 
> When hw.pci.enable_aspm is enabled, PCIe hot plug of Micron NVMe drives
> does not work (including detection on boot) and the system reports the
> following errors for the vast majority of the drives:
> 
> Nov 13 15:11:30 xxx kernel: pcib9: Timed out waiting for Data Link Layer
> Active
> Nov 13 15:12:17 xxx kernel: pcib30: Timed out waiting for Data Link
> Layer Active
> Nov 13 15:12:44 xxx kernel: pcib29: Timed out waiting for Data Link
> Layer Active
> Nov 13 15:13:04 xxx kernel: pcib11: Timed out waiting for Data Link
> Layer Active
> ...
> 
> Flipping the tunable back to 0 workarounds the problem.
> 
> Reproduced both on latest head and stable/12.

-- 
Alexander Motin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?e0dd58bc-251b-b80c-b3b6-e4ebbbc00229>