Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:1 as permitted sender) client-ip=2620:137:e000::3:1;
Date:   Thu, 7 Dec 2023 15:55:41 +0100
From:   Igor Mammedov <imammedo@redhat.com>
To:     Fiona Ebner <f.ebner@proxmox.com>
Cc:     Bjorn Helgaas <helgaas@kernel.org>, linux-pci@vger.kernel.org,
        linux-acpi@vger.kernel.org, linux-kernel@vger.kernel.org,
        bhelgaas@google.com, lenb@kernel.org, rafael@kernel.org,
        Thomas Lamprecht <t.lamprecht@proxmox.com>
Subject: Re: SCSI hotplug issues with UEFI VM with guest kernel >= 6.5
Message-ID: <20231207155541.735e0055@imammedo.users.ipa.redhat.com>
In-Reply-To: <4dbc72ba-8edb-4ff5-b95d-b601189e4415@proxmox.com>
References: <20231130231802.GA498017@bhelgaas>
        <4dbc72ba-8edb-4ff5-b95d-b601189e4415@proxmox.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Fri, 1 Dec 2023 10:24:41 +0100
Fiona Ebner <f.ebner@proxmox.com> wrote:

> Am 01.12.23 um 00:18 schrieb Bjorn Helgaas:
> > On Wed, Nov 29, 2023 at 04:22:41PM +0100, Fiona Ebner wrote:  
> >> Hi,
> >> it seems that hot-plugging SCSI disks for QEMU virtual machines booting
> >> with UEFI and with guest kernels >= 6.5 might be broken. It's not
> >> consistently broken, hinting there might be a race somewhere.
> >>
> >> Reverting the following two commits seems to make it work reliably again:
> >>
> >> cc22522fd55e2 ("PCI: acpiphp: Use
> >> pci_assign_unassigned_bridge_resources() only for non-root bus")
> >> 40613da52b13f ("PCI: acpiphp: Reassign resources on bridge if necessary"
> >>
> >> Of course, they might only expose some pre-existing issue, but this is
> >> my best lead. See below for some logs and details about an affected
> >> virtual machine. Happy to provide more information and to debug/test
> >> further.  
> > 
> > Shoot.  Thanks very much for the report and your debugging.  I'm
> > hoping Igor will chime in with some ideas.
> > 
> > Both of those commits appeard in v6.5 and fixed legit issues, so I
> > hate to revert them, but this does appear to be a regression.
> > 
> > #regzbot introduced: cc22522fd55e2 ^
> > #regzbot introduced: 40613da52b13f ^
> >   
> >> Host kernel: 6.5.11-4-pve which is based on the one from Ubuntu
> >> Guest kernel: 6.7.0-rc3 and 6.7.0-rc3 with above commits reverted
> >> QEMU version: v8.1.0 built from source
> >> EDK2 version: submodule in the QEMU v8.1 repository: edk2-stable202302
> >>  
> 
> I should mention that I haven't run into the issue when booting the VM
> with SeaBIOS yet.
> 
> Log for 6.7.0-rc3 + SeaBIOS (bundled with QEMU 8.1):
> 
> > Dec 01 10:08:08 hotplug kernel: pci 0000:01:02.0: [1af4:1004] type 00 class 0x010000
> > Dec 01 10:08:08 hotplug kernel: pci 0000:01:02.0: reg 0x10: [io  0x0000-0x003f]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:01:02.0: reg 0x14: [mem 0x00000000-0x00000fff]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:01:02.0: reg 0x20: [mem 0x00000000-0x00003fff 64bit pref]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:01:02.0: BAR 4: assigned [mem 0xfd404000-0xfd407fff 64bit pref]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:01:02.0: BAR 1: assigned [mem 0xfe801000-0xfe801fff]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:01:02.0: BAR 0: assigned [io  0xe040-0xe07f]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:00:05.0: PCI bridge to [bus 01]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:00:05.0:   bridge window [io  0xe000-0xefff]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:00:05.0:   bridge window [mem 0xfe800000-0xfe9fffff]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:00:05.0:   bridge window [mem 0xfd400000-0xfd5fffff 64bit pref]
> > Dec 01 10:08:08 hotplug kernel: virtio-pci 0000:01:02.0: enabling device (0000 -> 0003)
> > Dec 01 10:08:08 hotplug kernel: ACPI: \_SB_.LNKC: Enabled at IRQ 11
> > Dec 01 10:08:08 hotplug kernel: scsi host3: Virtio SCSI HBA
> > Dec 01 10:08:08 hotplug kernel: pci 0000:00:05.0: PCI bridge to [bus 01]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:00:05.0:   bridge window [io  0xe000-0xefff]
> > Dec 01 10:08:08 hotplug kernel: pci 0000:00:05.0:   bridge window [mem 0xfe800000-0xfe9fffff]
> > Dec 01 10:08:08 hotplug kernel: scsi 3:0:0:1: Direct-Access     QEMU     QEMU HARDDISK    2.5+ PQ: 0 ANSI: 5
> > Dec 01 10:08:08 hotplug kernel: pci 0000:00:05.0:   bridge window [mem 0xfd400000-0xfd5fffff 64bit pref]
> > Dec 01 10:08:08 hotplug kernel: scsi 3:0:0:1: Attached scsi generic sg1 type 0
> > Dec 01 10:08:08 hotplug kernel: sd 3:0:0:1: Power-on or device reset occurred
> > Dec 01 10:08:08 hotplug kernel: sd 3:0:0:1: [sdb] 2048 512-byte logical blocks: (1.05 MB/1.00 MiB)
> > Dec 01 10:08:08 hotplug kernel: sd 3:0:0:1: [sdb] Write Protect is off
> > Dec 01 10:08:08 hotplug kernel: sd 3:0:0:1: [sdb] Mode Sense: 63 00 00 08
> > Dec 01 10:08:08 hotplug kernel: sd 3:0:0:1: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
> > Dec 01 10:08:08 hotplug kernel: sd 3:0:0:1: [sdb] Attached SCSI disk  
> 
> Interestingly, the line with "QEMU HARDDISK" does not come after all
> lines with "bridge window" like was the case for the one time it did
> work with UEFI. So maybe that was just a red herring.

I've just seen this,
let me try to reproduce and see what can be done with it.

> 
> Best Regards,
> Fiona
>