Re: [PATCH] pci: Prevent VPD access for QLogic ISP2722

From: Ethan Zhao
Date: Tue Mar 07 2017 - 21:19:52 EST




On 2017/3/8 4:22, Bjorn Helgaas wrote:
On Mon, Feb 27, 2017 at 05:08:44PM +0900, Ethan Zhao wrote:
QLogic ISP2722-based 16/32Gb Fibre Channel to PCIe Adapter has the VPD
access issue too, while read the common pci-sysfs access interface shown as

/sys/devices/pci0000:00/0000:00:03.2/0000:0b:00.0/vpd

with simple 'cat' could cause system hang and panic:

[ 48.219586] Kernel panic - not syncing: An NMI occurred. Depending on your system
the reason for the NMI is logged in any one of the following resources:
[ 48.219586] 1. Integrated Management Log (IML)
[ 48.219586] 2. OA Syslog
[ 48.219586] 3. OA Forward Progress Log
[ 48.219586] 4. iLO Event Log
[ 48.352034] CPU: 0 PID: 15070 Comm: udevadm Not tainted 4.1.12
[ 48.390989] Hardware name: HP ProLiant DL380 Gen9/ProLiant DL380 Gen9, BIOS P89 12/27/2015
[ 48.431201] 0000000000000086 000000007f0cdf51 ffff880c4fa05d58 ffffffff817193de
[ 48.467185] ffffffffa00b42d8 0000000000000075 ffff880c4fa05dd8 ffffffff81714072
[ 48.502631] 0000000000000008 ffff880c4fa05de8 ffff880c4fa05d88 000000007f0cdf51
[ 48.538427] Call Trace:
[ 48.549991] <NMI> [<ffffffff817193de>] dump_stack+0x63/0x81
[ 48.577496] [<ffffffff81714072>] panic+0xd0/0x20e
[ 48.600598] [<ffffffffa00b390d>] hpwdt_pretimeout+0xdd/0xe0 [hpwdt]
[ 48.630913] [<ffffffff81021fc9>] ? sched_clock+0x9/0x10
[ 48.656286] [<ffffffff8101c101>] nmi_handle+0x91/0x170
[ 48.681362] [<ffffffff8101c10c>] ? nmi_handle+0x9c/0x170
[ 48.707383] [<ffffffff8101c5fe>] io_check_error+0x1e/0xa0
[ 48.733603] [<ffffffff8101c719>] default_do_nmi+0x99/0x140
[ 48.759570] [<ffffffff8101c8b4>] do_nmi+0xf4/0x170
[ 48.782977] [<ffffffff817232c5>] end_repeat_nmi+0x1a/0x1e
[ 48.809290] [<ffffffff815d724b>] ? pci_conf1_read+0xeb/0x120
[ 48.837782] [<ffffffff815d724b>] ? pci_conf1_read+0xeb/0x120
[ 48.865045] [<ffffffff815d724b>] ? pci_conf1_read+0xeb/0x120
[ 48.892877] <<EOE>> [<ffffffff815db4b3>] raw_pci_read+0x23/0x40
[ 48.923060] [<ffffffff815db4fc>] pci_read+0x2c/0x30
[ 48.947340] [<ffffffff8136f612>] pci_user_read_config_word+0x72/0x110
[ 48.978937] [<ffffffff8136f746>] pci_vpd_pci22_wait+0x96/0x130
[ 49.007983] [<ffffffff8136ff9b>] pci_vpd_pci22_read+0xdb/0x1a0
[ 49.036370] [<ffffffff8136ea30>] pci_read_vpd+0x20/0x30
[ 49.062608] [<ffffffff8137d590>] read_vpd_attr+0x30/0x40
[ 49.088627] [<ffffffff8128e037>] sysfs_kf_bin_read+0x47/0x70
[ 49.116383] [<ffffffff8128d24e>] kernfs_fop_read+0xae/0x180
[ 49.143396] [<ffffffff8120dd97>] __vfs_read+0x37/0x100
[ 49.169059] [<ffffffff812ba7e4>] ? security_file_permission+0x84/0xa0
[ 49.200526] [<ffffffff8120e366>] ? rw_verify_area+0x56/0xe0
[ 49.227578] [<ffffffff8120e476>] vfs_read+0x86/0x140
[ 49.252842] [<ffffffff8120f3f5>] SyS_read+0x55/0xd0
[ 49.277378] [<ffffffff81720f2e>] system_call_fastpath+0x12/0x71
[ 50.349812] Shutting down cpus with NMI
[ 50.368388] Kernel Offset: disabled
[ 50.385327] drm_kms_helper: panic occurred, switching back to text console

So blacklist the access to its VPD.

Signed-off-by: Ethan Zhao <ethan.zhao@xxxxxxxxxx>
Applied to for-linus for v4.11, thanks!

I added a stable tag for v4.6+ because quirk_blacklist_vpd() was added
by 7c20078a8197 ("PCI: Prevent VPD access for buggy devices"), which
appeared in v4.6.
Nice ! Thanks again.

Ethan
---
drivers/pci/quirks.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index 3a035e07..087a218 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -2173,6 +2173,7 @@ static void quirk_blacklist_vpd(struct pci_dev *dev)
DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_LSI_LOGIC, 0x005f, quirk_blacklist_vpd);
DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATTANSIC, PCI_ANY_ID,
quirk_blacklist_vpd);
+DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_QLOGIC, 0x2261, quirk_blacklist_vpd);
/*
* For Broadcom 5706, 5708, 5709 rev. A nics, any read beyond the
--
1.8.3.1