| 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
| ||
|
Message-ID: <17532a01-b4fd-4538-b0dd-d88d3ff30784@linux.ibm.com>
Date: Tue, 13 Aug 2024 14:49:29 -0400
From: Matthew Rosato <mjrosato@...ux.ibm.com>
To: Niklas Schnelle <schnelle@...ux.ibm.com>,
Yunsheng Lin <linyunsheng@...wei.com>,
Somnath Kotur <somnath.kotur@...adcom.com>,
Jesper Dangaard Brouer <hawk@...nel.org>
Cc: Yonglong Liu <liuyonglong@...wei.com>,
"David S. Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>, pabeni@...hat.com,
ilias.apalodimas@...aro.org, netdev@...r.kernel.org,
linux-kernel@...r.kernel.org,
Alexander Duyck <alexander.duyck@...il.com>,
Alexei Starovoitov <ast@...nel.org>,
"shenjian (K)" <shenjian15@...wei.com>,
Salil Mehta <salil.mehta@...wei.com>, joro@...tes.org, will@...nel.org,
robin.murphy@....com, iommu@...ts.linux.dev
Subject: Re: [BUG REPORT]net: page_pool: kernel crash at
iommu_get_dma_domain+0xc/0x20
On 8/6/24 9:35 AM, Niklas Schnelle wrote:
> On Mon, 2024-08-05 at 20:19 +0800, Yunsheng Lin wrote:
>> On 2024/7/31 16:42, Somnath Kotur wrote:
>>> On Tue, Jul 30, 2024 at 10:51 PM Jesper Dangaard Brouer <hawk@...nel.org> wrote:
>>>>
>>
>> +cc iommu maintainers and list
>>
>>>>
>>>> On 30/07/2024 15.08, Yonglong Liu wrote:
>>>>> I found a bug when running hns3 driver with page pool enabled, the log
>>>>> as below:
>>>>>
>>>>> [ 4406.956606] Unable to handle kernel NULL pointer dereference at
>>>>> virtual address 00000000000000a8
>>>>
>>>> struct iommu_domain *iommu_get_dma_domain(struct device *dev)
>>>> {
>>>> return dev->iommu_group->default_domain;
>>>> }
>>>>
>>>> $ pahole -C iommu_group --hex | grep default_domain
>>>> struct iommu_domain * default_domain; /* 0xa8 0x8 */
>>>>
>>>> Looks like iommu_group is a NULL pointer (that when deref member
>>>> 'default_domain' cause this fault).
>>>>
>>>>
>>>>> [ 4406.965379] Mem abort info:
>>>>> [ 4406.968160] ESR = 0x0000000096000004
>>>>> [ 4406.971906] EC = 0x25: DABT (current EL), IL = 32 bits
>>>>> [ 4406.977218] SET = 0, FnV = 0
>>>>> [ 4406.980258] EA = 0, S1PTW = 0
>>>>> [ 4406.983404] FSC = 0x04: level 0 translation fault
>>>>> [ 4406.988273] Data abort info:
>>>>> [ 4406.991154] ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
>>>>> [ 4406.996632] CM = 0, WnR = 0, TnD = 0, TagAccess = 0
>>>>> [ 4407.001681] GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
>>>>> [ 4407.006985] user pgtable: 4k pages, 48-bit VAs, pgdp=0000202828326000
>>>>> [ 4407.013430] [00000000000000a8] pgd=0000000000000000,
>>>>> p4d=0000000000000000
>>>>> [ 4407.020212] Internal error: Oops: 0000000096000004 [#1] PREEMPT SMP
>>>>> [ 4407.026454] Modules linked in: hclgevf xt_CHECKSUM ipt_REJECT
>>>>> nf_reject_ipv4 ip6table_mangle ip6table_nat iptable_mangle
>>>>> ip6table_filter ip6_tables hns_roce_hw_v2 hns3 hclge hnae3 xt_addrtype
>>>>> iptable_filter xt_conntrack overlay arm_spe_pmu arm_smmuv3_pmu
>>>>> hisi_uncore_hha_pmu hisi_uncore_ddrc_pmu hisi_uncore_l3c_pmu
>>>>> hisi_uncore_pmu fuse rpcrdma ib_isert iscsi_target_mod ib_iser libiscsi
>>>>> scsi_transport_iscsi crct10dif_ce hisi_sec2 hisi_hpre hisi_zip
>>>>> hisi_sas_v3_hw xhci_pci sbsa_gwdt hisi_qm hisi_sas_main hisi_dma
>>>>> xhci_pci_renesas uacce libsas [last unloaded: hnae3]
>>>>> [ 4407.076027] CPU: 48 PID: 610 Comm: kworker/48:1
>>>>> [ 4407.093343] Workqueue: events page_pool_release_retry
>>>>> [ 4407.098384] pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS
>>>>> BTYPE=--)
>>>>> [ 4407.105316] pc : iommu_get_dma_domain+0xc/0x20
>>>>> [ 4407.109744] lr : iommu_dma_unmap_page+0x38/0xe8
>>>>> [ 4407.114255] sp : ffff80008bacbc80
>>>>> [ 4407.117554] x29: ffff80008bacbc80 x28: 0000000000000000 x27:
>>>>> ffffc31806be7000
>>>>> [ 4407.124659] x26: ffff2020002b6ac0 x25: 0000000000000000 x24:
>>>>> 0000000000000002
>>>>> [ 4407.131762] x23: 0000000000000022 x22: 0000000000001000 x21:
>>>>> 00000000fcd7c000
>>>>> [ 4407.138865] x20: ffff0020c9882800 x19: ffff0020856f60c8 x18:
>>>>> ffff8000d3503c58
>>>>> [ 4407.145968] x17: 0000000000000000 x16: 1fffe00419521061 x15:
>>>>> 0000000000000001
>>>>> [ 4407.153073] x14: 0000000000000003 x13: 00000401850ae012 x12:
>>>>> 000006b10004e7fb
>>>>> [ 4407.160177] x11: 0000000000000067 x10: 0000000000000c70 x9 :
>>>>> ffffc3180405cd20
>>>>> [ 4407.167280] x8 : fefefefefefefeff x7 : 0000000000000001 x6 :
>>>>> 0000000000000010
>>>>> [ 4407.174382] x5 : ffffc3180405cce8 x4 : 0000000000000022 x3 :
>>>>> 0000000000000002
>>>>> [ 4407.181485] x2 : 0000000000001000 x1 : 00000000fcd7c000 x0 :
>>>>> 0000000000000000
>>>>> [ 4407.188589] Call trace:
>>>>> [ 4407.191027] iommu_get_dma_domain+0xc/0x20
>>>>> [ 4407.195105] dma_unmap_page_attrs+0x38/0x1d0
>>>>> [ 4407.199361] page_pool_return_page+0x48/0x180
>>>>> [ 4407.203699] page_pool_release+0xd4/0x1f0
>>>>> [ 4407.207692] page_pool_release_retry+0x28/0xe8
>>>>
>>>> I suspect that the DMA IOMMU part was deallocated and freed by the
>>>> driver even-though page_pool still have inflight packets.
>>> When you say driver, which 'driver' do you mean?
>>> I suspect this could be because of the VF instance going away with
>>> this cmd - disable the vf: echo 0 >
>>> /sys/class/net/eno1/device/sriov_numvfs, what do you think?
>>>>
>>>> The page_pool bumps refcnt via get_device() + put_device() on the DMA
>>>> 'struct device', to avoid it going away, but I guess there is also some
>>>> IOMMU code that we need to make sure doesn't go away (until all inflight
>>>> pages are returned) ???
>>
>> I guess the above is why thing went wrong here, the question is which
>> IOMMU code need to be called here to stop them from going away.
>>
>> What I am also curious is that there should be a pool of allocated iova in
>> iommu that is corresponding to the in-flight page for page_pool, shouldn't
>> iommu wait for the corresponding allocated iova to be freed similarly as
>> page_pool does for it's in-flight pages?
>>
>
>
> Is it possible you're using an IOMMU whose driver doesn't yet support
> blocking_domain? I'm currently working an issue on s390 that also
> occurs during device removal and is fixed by implementing blocking
> domain in the s390 IOMMU driver (patch forthcoming). The root cause for
> that is that our domain->ops->attach_dev() fails when during hot-unplug
> the device is already gone from the platform's point of view and then
> we ended up with a NULL domain unless we have a blocking domain which
> can handle non existant devices and gets set as fallback in
> __iommu_device_set_domain(). In the case I can reproduce the backtrace
> is different[0] but we also saw at least two cases where we see the
> exact same call trace as in the first mail of this thread. So far I
> suspected them to be due to the blocking domain issue but it could be a
> separate issue too.
>
> Thanks,
> Niklas
>
Couple of things to follow up with on Niklas' statement above...
So first, after further testing on my end, I wanted to clarify that the implementation of a blocked domain is unrelated to this bug. Sorry for the noise.
Second, it looks like Niklas copied an unrelated backtrace in his report (I snipped it from my reply).
But I wanted to be clear, we can reproduce this same sort of error on s390 and using a different device driver (mlx5_core) and the backtrace is almost identical to what this thread is reporting and in the same area. The most reliable repro method I've found so far is to use a few mellanox VFs and power one down (echo 0 > /sys/bus/pci/slots/.../power) during or shortly after a tcp workload (iperf3). I verified that I can reproduce on a kernel as old as 6.7 release tag; I stopped there because that's the release when s390 converted to use dma-iommu but I assume the problem really existed longer than this.
Here's a backtrace of a repro on s390 (6.11-rc3):
[ 691.860855] Unable to handle kernel pointer dereference in virtual kernel address space
[ 691.861089] Failing address: 0706c00180000000 TEID: 0706c00180000803
[ 691.861097] Fault in home space mode while using kernel ASCE.
[ 691.861118] AS:0000000154fbc007 R3:0000000000000024
[ 691.861153] Oops: 0038 ilc:2 [#1] PREEMPT SMP
[ 691.861161] Modules linked in: xt_CHECKSUM xt_MASQUERADE xt_conntrack ipt_REJECT nf_reject_ipv4 xt_tcpudp ip6table_mangle ip6table_nat ip6table_filter ip6_tables iptable_mangle iptable_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 iptable_filter ip_tables x_tables rpcrdma rdma_ucm rdma_cm iw_cm ib_cm uvdevic
e s390_trng ism eadm_sch sunrpc tape_34xx tape tape_class mlx5_ib vfio_ap ib_uverbs kvm ib_core zcrypt_cex4 vfio_ccw mdev vfio_iommu_type1 vfio sch_fq_codel loop dm_multipath nfnetlink lcs ctcm fsm mlx5_core ghash_s390 prng chacha_s390 aes_s390 des_s390 libdes sha3_512_s390 sha3_256_s390 sha512_s390 sha256_s390 sha1_s3
90 sha_common zfcp scsi_transport_fc scsi_dh_rdac scsi_dh_emc scsi_dh_alua pkey zcrypt rng_core autofs4
[ 691.861319] CPU: 9 UID: 0 PID: 283 Comm: kworker/9:2 Not tainted 6.11.0-rc3 #1
[ 691.861325] Hardware name: IBM 8561 T01 772 (LPAR)
[ 691.861329] Workqueue: events page_pool_release_retry
[ 691.861342] Krnl PSW : 0704e00180000000 0000013bd3a6ee26 (iommu_iova_to_phys+0x6/0x40)
[ 691.861355] R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:2 PM:0 RI:0 EA:3
[ 691.861363] Krnl GPRS: 0000000080000000 0000000000000000 0706c00180000000 0000000afba90000
[ 691.861369] 0000000000001000 0000000000000002 0000000000000022 0000000000000002
[ 691.861374] 0000000000001000 0000000afba90000 00000039c0e19000 00000039c0bc9098
[ 691.861379] 000000427ac33400 00000039c0e19068 0000013bd3a7655c 000000bbd8d1bb28
[ 691.861389] Krnl Code: 0000013bd3a6ee1c: 0707 bcr 0,%r7
[ 691.861389] 0000013bd3a6ee1e: 0707 bcr 0,%r7
[ 691.861389] #0000013bd3a6ee20: c004002a8d6c brcl 0,0000013bd3fc08f8
[ 691.861389] >0000013bd3a6ee26: 58502000 l %r5,0(%r2)
[ 691.861389] 0000013bd3a6ee2a: ec580014047e cij %r5,4,8,0000013bd3a6ee52
[ 691.861389] 0000013bd3a6ee30: ec58000c007e cij %r5,0,8,0000013bd3a6ee48
[ 691.861389] 0000013bd3a6ee36: e31020080004 lg %r1,8(%r2)
[ 691.861389] 0000013bd3a6ee3c: e31010400004 lg %r1,64(%r1)
[ 691.861426] Call Trace:
[ 691.861430] [<0000013bd3a6ee26>] iommu_iova_to_phys+0x6/0x40
[ 691.861436] [<0000013bd2f47a32>] dma_unmap_page_attrs+0x1a2/0x1e0
[ 691.861443] [<0000013bd3c2b81a>] page_pool_return_page+0x5a/0x130
[ 691.861449] [<0000013bd3c2cb68>] page_pool_release+0xb8/0x1f0
[ 691.861455] [<0000013bd3c2ce9c>] page_pool_release_retry+0x2c/0x120
[ 691.861461] [<0000013bd2e95652>] process_one_work+0x2b2/0x5d0
[ 691.861467] [<0000013bd2e9625e>] worker_thread+0x20e/0x3f0
[ 691.861473] [<0000013bd2ea25e2>] kthread+0x152/0x170
[ 691.861478] [<0000013bd2e135ac>] __ret_from_fork+0x3c/0x60
[ 691.861484] [<0000013bd3ecf0ca>] ret_from_fork+0xa/0x38
[ 691.861491] INFO: lockdep is turned off.
[ 691.861495] Last Breaking-Event-Address:
[ 691.861499] [<0000013bd3a76556>] iommu_dma_unmap_page+0x36/0xb0
[ 691.861507] Kernel panic - not syncing: Fatal exception: panic_on_oops
Powered by blists - more mailing lists