Re: [PATCH net] net: hns3: Clear the CMDQ registers before unmapping BAR region

From: Yunsheng Lin
Date: Fri Oct 23 2020 - 02:22:04 EST


On 2020/10/23 13:15, Zenghui Yu wrote:
> When unbinding the hns3 driver with the HNS3 VF, I got the following
> kernel panic:
>
> [ 265.709989] Unable to handle kernel paging request at virtual address ffff800054627000
> [ 265.717928] Mem abort info:
> [ 265.720740] ESR = 0x96000047
> [ 265.723810] EC = 0x25: DABT (current EL), IL = 32 bits
> [ 265.729126] SET = 0, FnV = 0
> [ 265.732195] EA = 0, S1PTW = 0
> [ 265.735351] Data abort info:
> [ 265.738227] ISV = 0, ISS = 0x00000047
> [ 265.742071] CM = 0, WnR = 1
> [ 265.745055] swapper pgtable: 4k pages, 48-bit VAs, pgdp=0000000009b54000
> [ 265.751753] [ffff800054627000] pgd=0000202ffffff003, p4d=0000202ffffff003, pud=00002020020eb003, pmd=00000020a0dfc003, pte=0000000000000000
> [ 265.764314] Internal error: Oops: 96000047 [#1] SMP
> [ 265.830357] CPU: 61 PID: 20319 Comm: bash Not tainted 5.9.0+ #206
> [ 265.836423] Hardware name: Huawei TaiShan 2280 V2/BC82AMDDA, BIOS 1.05 09/18/2019
> [ 265.843873] pstate: 80400009 (Nzcv daif +PAN -UAO -TCO BTYPE=--)
> [ 265.843890] pc : hclgevf_cmd_uninit+0xbc/0x300
> [ 265.861988] lr : hclgevf_cmd_uninit+0xb0/0x300
> [ 265.861992] sp : ffff80004c983b50
> [ 265.881411] pmr_save: 000000e0
> [ 265.884453] x29: ffff80004c983b50 x28: ffff20280bbce500
> [ 265.889744] x27: 0000000000000000 x26: 0000000000000000
> [ 265.895034] x25: ffff800011a1f000 x24: ffff800011a1fe90
> [ 265.900325] x23: ffff0020ce9b00d8 x22: ffff0020ce9b0150
> [ 265.905616] x21: ffff800010d70e90 x20: ffff800010d70e90
> [ 265.910906] x19: ffff0020ce9b0080 x18: 0000000000000004
> [ 265.916198] x17: 0000000000000000 x16: ffff800011ae32e8
> [ 265.916201] x15: 0000000000000028 x14: 0000000000000002
> [ 265.916204] x13: ffff800011ae32e8 x12: 0000000000012ad8
> [ 265.946619] x11: ffff80004c983b50 x10: 0000000000000000
> [ 265.951911] x9 : ffff8000115d0888 x8 : 0000000000000000
> [ 265.951914] x7 : ffff800011890b20 x6 : c0000000ffff7fff
> [ 265.951917] x5 : ffff80004c983930 x4 : 0000000000000001
> [ 265.951919] x3 : ffffa027eec1b000 x2 : 2b78ccbbff369100
> [ 265.964487] x1 : 0000000000000000 x0 : ffff800054627000
> [ 265.964491] Call trace:
> [ 265.964494] hclgevf_cmd_uninit+0xbc/0x300
> [ 265.964496] hclgevf_uninit_ae_dev+0x9c/0xe8
> [ 265.964501] hnae3_unregister_ae_dev+0xb0/0x130
> [ 265.964516] hns3_remove+0x34/0x88 [hns3]
> [ 266.009683] pci_device_remove+0x48/0xf0
> [ 266.009692] device_release_driver_internal+0x114/0x1e8
> [ 266.030058] device_driver_detach+0x28/0x38
> [ 266.034224] unbind_store+0xd4/0x108
> [ 266.037784] drv_attr_store+0x40/0x58
> [ 266.041435] sysfs_kf_write+0x54/0x80
> [ 266.045081] kernfs_fop_write+0x12c/0x250
> [ 266.049076] vfs_write+0xc4/0x248
> [ 266.052378] ksys_write+0x74/0xf8
> [ 266.055677] __arm64_sys_write+0x24/0x30
> [ 266.059584] el0_svc_common.constprop.3+0x84/0x270
> [ 266.064354] do_el0_svc+0x34/0xa0
> [ 266.067658] el0_svc+0x38/0x40
> [ 266.070700] el0_sync_handler+0x8c/0xb0
> [ 266.074519] el0_sync+0x140/0x180
>
> It looks like the BAR memory region had already been unmapped before we
> start clearing CMDQ registers in it, which is pretty bad and the kernel
> happily kills itself because of a Current EL Data Abort (on arm64).
>
> Moving the CMDQ uninitialization a bit early fixes the issue for me.

Yes, this seems a obvious bug, and below patch seems the obvious fix too.

We has testcase to test the loading and unloading process, but does not
seem to catch this error, and testing it again using the latest internal
version seems ok too:

[root@localhost ~]# rmmod hclgevf
[ 8035.010715] hns3 0000:7d:01.0 eth8: net stop
[ 8035.079917] hns3 0000:7d:01.0 eth8: link down
[ 8036.402491] hns3 0000:7d:01.1 eth9: net stop
[ 8036.472946] hns3 0000:7d:01.1 eth9: link down
[root@localhost ~]# rmmod hns3
[ 8045.938705] hns3 0000:bd:00.3 eth7: net stop
[ 8046.354308] hns3 0000:bd:00.2 eth6: net stop
[ 8046.627653] hns3 0000:bd:00.1 eth5: net stop
[ 8046.632251] hns3 0000:bd:00.1 eth5: link down
[ 8047.050235] hns3 0000:bd:00.0 eth4: net stop
[ 8047.054837] hns3 0000:bd:00.0 eth4: link down
[ 8047.340528] hns3 0000:7d:00.3 eth3: net stop
[ 8047.347633] hns3 0000:7d:00.3 eth3: link down
[ 8048.879299] hns3 0000:7d:00.2 eth2: net stop
[ 8050.271999] hns3 0000:7d:00.1 eth1: net stop
[ 8050.278755] hns3 0000:7d:00.1 eth1: link down
[ 8051.650607] pci 0000:7d:01.0: Removing from iommu group 44
[ 8051.657909] pci 0000:7d:01.1: Removing from iommu group 45
[ 8052.794671] hns3 0000:7d:00.0 eth0: net stop
[ 8052.800385] hns3 0000:7d:00.0 eth0: link down
[root@localhost ~]#
[root@localhost ~]#


Do you care to provide the testcase for above calltrace?

>
> Signed-off-by: Zenghui Yu <yuzenghui@xxxxxxxxxx>
> ---
>
> I have almost zero knowledge about the hns3 driver. You can regard this
> as a report and make a better fix if possible.
>
> I can't even figure out that how can we live with this issue for a long
> time... It should exists since commit 34f81f049e35 ("net: hns3: clear
> command queue's registers when unloading VF driver"), where we start
> writing something into the unmapped area.
>
> drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
> index 50c84c5e65d2..c8e3fdd5999c 100644
> --- a/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
> +++ b/drivers/net/ethernet/hisilicon/hns3/hns3vf/hclgevf_main.c
> @@ -3262,8 +3262,8 @@ static void hclgevf_uninit_hdev(struct hclgevf_dev *hdev)
> hclgevf_uninit_msi(hdev);
> }
>
> - hclgevf_pci_uninit(hdev);
> hclgevf_cmd_uninit(hdev);
> + hclgevf_pci_uninit(hdev);
> hclgevf_uninit_mac_list(hdev);
> }
>
>