linux-kernel - Re: [Bug] Driver mt7921e cause computer reboot.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20211230002118.GA1717394@bhelgaas>
Date:   Wed, 29 Dec 2021 18:21:18 -0600
From:   Bjorn Helgaas <helgaas@...nel.org>
To:     Mikhail Gavrilov <mikhail.v.gavrilov@...il.com>
Cc:     Íñigo Huguet <ihuguet@...hat.com>,
        Linux List Kernel Mailing <linux-kernel@...r.kernel.org>,
        sean.wang@...iatek.com,
        Linux List Kernel Mailing <linux-wireless@...r.kernel.org>,
        objelf@...il.com, lorenzo@...nel.org, Soul.Huang@...iatek.com,
        Felix Fietkau <nbd@....name>, linux-pci@...r.kernel.org,
        Lorenzo Bianconi <lorenzo.bianconi83@...il.com>,
        Ryder Lee <ryder.lee@...iatek.com>
Subject: Re: [Bug] Driver mt7921e cause computer reboot.

[+cc Lorenzo, Ryder (the rest of the mt7921 maintainers)]

Thread: https://lore.kernel.org/all/CABXGCsODP8ze_mvzfJKcRYxuS-esVgHXAvDXS5KN3xFUN6bWgA@mail.gmail.com/T/#u

On Mon, Dec 27, 2021 at 04:30:11PM +0500, Mikhail Gavrilov wrote:
> On Mon, 27 Dec 2021 at 15:11, Íñigo Huguet <ihuguet@...hat.com> wrote:
> > I've been experiencing similar problems, but they're solved at v5.15
> > version, at least for me.
> >
> > How are you installing the kernel? Custom build? Have you updated the
> > firmware to latest versions, as well?
> 
> I use Fedora Rawhide with default kernel and firmware packages.
> 
> $ uname -r
> 5.16.0-0.rc6.20211223gitbc491fb12513.44.fc36.x86_64
> $ rpm -q linux-firmware
> linux-firmware-20211027-126.fc36.noarch
> 
> >
> > For me, these differences seem to be the normal effect of the driver
> > not recognizing the device.
> 
> By the kernel logs, it looks like this:
> After reboot:
> $ dmesg | grep mt7921e
> [    8.629358] mt7921e 0000:05:00.0: enabling device (0000 -> 0002)
> [    8.630229] mt7921e 0000:05:00.0: ASIC revision: 79610010
> [    9.687652] mt7921e: probe of 0000:05:00.0 failed with error -110
> 
> # rmmod mt7921e
> # modprobe mt7921e
> 
> [  215.514503] mt7921e 0000:05:00.0: ASIC revision: feed0000
> [  216.604741] mt7921e: probe of 0000:05:00.0 failed with error -110
> 
> After cold boot after shutdown:
> $ dmesg | grep mt7921e
> [    8.545171] mt7921e 0000:05:00.0: enabling device (0000 -> 0002)
> [    8.545757] mt7921e 0000:05:00.0: ASIC revision: 79610010
> [    8.631156] mt7921e 0000:05:00.0: HW/SW Version: 0x8a108a10, Build
> Time: 20211014150838a
> [    8.912687] mt7921e 0000:05:00.0: WM Firmware Version: ____010000,
> Build Time: 20211014150922
> [    8.938756] mt7921e 0000:05:00.0: Firmware init done
> [    9.753257] mt7921e 0000:05:00.0 wlp5s0: renamed from wlan0
> 
> It looks like something is not re-initialized after a reboot.
> Laptop BIOS is latest: Version 316
> https://dlcdnets.asus.com/pub/ASUS/GamingNB/G513QY/G513QYAS316.zip
> 
> Maybe anyone from the pci mailing list can lid some light why pci
> device not re-initialized after a reboot?

Sorry for the inconvenience and thank you very much for the report!

If I understand correctly, when you do a cold boot, the mt7921e device
works properly.

But when you simply reboot, without a power off, the device does not
work, and the dmesg log contains:

  pci 0000:05:00.0: [14c3:7961] type 00 class 0x028000
  pci 0000:05:00.0: reg 0x10: [mem 0xfc30300000-0xfc303fffff 64bit pref]
  pci 0000:05:00.0: reg 0x18: [mem 0xfc30400000-0xfc30403fff 64bit pref]
  pci 0000:05:00.0: reg 0x20: [mem 0xfc30404000-0xfc30404fff 64bit pref]
  ...
  mt7921e 0000:05:00.0: enabling device (0000 -> 0002)
  mt7921e 0000:05:00.0: ASIC revision: 79610010
  mt7921e: probe of 0000:05:00.0 failed with error -110

That means the device responds to PCI config reads and writes, but the
probe failed with -ETIMEDOUT after printing the ASIC revision [1].

devm_request_irq() should not return -ETIMEDOUT, but it looks like
mt7921_dma_init() can (via mt7921_dma_disable()).  Maybe the mt7921e
driver can't tolerate some state the device was left in by reboot?

I don't see anything obviously wrong from a PCI core perspective.  The
PCI core does not reset devices either when going down for a reboot or
when coming up at boot-time.

Bjorn

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/net/wireless/mediatek/mt76/mt7921/pci.c?id=v5.16-rc6#n187