Message-ID: <20220930225639.q4hr4vcqhy7zyomk@skbuf>
Date:   Fri, 30 Sep 2022 22:56:40 +0000
From:   Vladimir Oltean <vladimir.oltean@....com>
To:     Vinicius Costa Gomes <vinicius.gomes@...el.com>
CC:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "yannick.vignon@....com" <yannick.vignon@....com>,
        "jhs@...atatu.com" <jhs@...atatu.com>,
        "xiyou.wangcong@...il.com" <xiyou.wangcong@...il.com>,
        "kuba@...nel.org" <kuba@...nel.org>,
        Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@...el.com>
Subject: Re: [PATCH net v1] net/sched: taprio: Fix crash when adding child
 qdisc

On Fri, Sep 30, 2022 at 03:30:42PM -0700, Vinicius Costa Gomes wrote:
> It was reported by Husaini that since commit 1461d212ab27 ("net/sched:
> taprio: make qdisc_leaf() see the per-netdev-queue pfifo child
> qdiscs"), there was a NULL dereference when adding a child qdisc (ETF
> in this case) when using taprio full offload.
> 
> The issue seems to be that since commit 13511704f8d7 ("net: taprio
> offload: enforce qdisc to netdev queue mapping"), the array of child
> qdiscs (q->qdiscs[]) was not used, and even free'd when full offload
> mode is used, in attach().
> 
> The current implementation since 1461d212ab27 (which I feel is a
> sensible change) requires that the q->qdiscs[] array is always valid.
> The fix is to not deallocate it, and keep it having references to each
> child qdisc.
> 
> Log:
> 
> [ 2295.318620] IPv6: ADDRCONF(NETDEV_CHANGE): enp170s0: link becomes ready
> [ 2371.758745] BUG: kernel NULL pointer dereference, address: 0000000000000008
> [ 2371.765697] #PF: supervisor read access in kernel mode
> [ 2371.770825] #PF: error_code(0x0000) - not-present page
> [ 2371.775952] PGD 0 P4D 0
> [ 2371.778494] Oops: 0000 [#1] PREEMPT SMP NOPTI
> [ 2371.782844] CPU: 0 PID: 17964 Comm: tc Tainted: G     U  W          6.0.0-rc6-intel-ese-standard-lts+ #95
> [ 2371.792374] Hardware name: Intel Corporation Tiger Lake Client Platform/TigerLake U DDR4 SODIMM RVP, BIOS TGLIFUI1.R00.4204.A00.2105270302 05/27/2021
> [ 2371.805701] RIP: 0010:taprio_leaf+0x1e/0x30 [sch_taprio]
> [ 2371.811011] Code: Unable to access opcode bytes at RIP 0xffffffffc1112ff4.
> [ 2371.817867] RSP: 0018:ffffae7f82263a28 EFLAGS: 00010206
> [ 2371.823078] RAX: 0000000000000000 RBX: 0000000001000002 RCX: ffff8a704ff38000
> [ 2371.830190] RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff8a7043f29000
> [ 2371.837307] RBP: ffff8a7043f29000 R08: 0000000000000000 R09: ffffae7f82263b98
> [ 2371.844420] R10: 0000000000000002 R11: ffff8a7045df0000 R12: ffff8a70519cc400
> [ 2371.851534] R13: ffffae7f82263b98 R14: ffffffff9a04fb00 R15: ffff8a704ff38000
> [ 2371.858646] FS:  00007fd2fc262740(0000) GS:ffff8a7790a00000(0000) knlGS:0000000000000000
> [ 2371.866707] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [ 2371.872435] CR2: ffffffffc1112ff4 CR3: 00000001053ac003 CR4: 0000000000770ef0
> [ 2371.879546] PKRU: 55555554
> [ 2371.882261] Call Trace:
> [ 2371.884719]  <TASK>
> [ 2371.886831]  tc_modify_qdisc+0x75/0x7c0
> [ 2371.890667]  rtnetlink_rcv_msg+0x141/0x3c0
> [ 2371.894764]  ? _copy_to_iter+0x1b0/0x5a0
> [ 2371.898685]  ? rtnl_calcit.isra.0+0x140/0x140
> [ 2371.903040]  netlink_rcv_skb+0x4e/0x100
> [ 2371.906878]  netlink_unicast+0x197/0x240
> [ 2371.910797]  netlink_sendmsg+0x246/0x4a0
> [ 2371.914718]  sock_sendmsg+0x5f/0x70
> [ 2371.918207]  ____sys_sendmsg+0x20f/0x280
> [ 2371.922130]  ? copy_msghdr_from_user+0x72/0xb0
> [ 2371.926570]  ___sys_sendmsg+0x7c/0xc0
> [ 2371.930233]  ? ___sys_recvmsg+0x89/0xc0
> [ 2371.934070]  __sys_sendmsg+0x59/0xa0
> [ 2371.937647]  do_syscall_64+0x40/0x90
> [ 2371.941224]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
> [ 2371.946273] RIP: 0033:0x7fd2fc38e707
> [ 2371.949847] Code: 0e 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b9 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10
> [ 2371.968521] RSP: 002b:00007fffafdf9ce8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
> [ 2371.976070] RAX: ffffffffffffffda RBX: 0000000063359bf0 RCX: 00007fd2fc38e707
> [ 2371.983183] RDX: 0000000000000000 RSI: 00007fffafdf9d50 RDI: 0000000000000003
> [ 2371.990293] RBP: 0000000000000000 R08: 0000000000000001 R09: 0000000000000000
> [ 2371.997406] R10: 00007fd2fc28d670 R11: 0000000000000246 R12: 0000000000000001
> [ 2372.004520] R13: 00005583e730c476 R14: 00005583e730c48a R15: 00005583e7333280
> [ 2372.011637]  </TASK>
> [ 2372.013830] Modules linked in: sch_taprio bnep 8021q bluetooth ecdh_generic ecc ecryptfs nfsd sch_fq_codel uio uhid iwlwifi cfg80211 i915 x86_pkg_temp_thermal kvm_intel kvm hid_sensor_als hid_sensor_incl_3d hid_sensor_gyro_3d hid_sensor_accel_3d hid_sensor_magn_3d hid_sensor_trigger hid_sensor_iio_common hid_sensor_custom hid_sensor_hub atkbd mei_hdcp intel_ishtp_hid libps2 mei_wdt vivaldi_fmap dwc3 dwmac_intel stmmac e1000e ax88179_178a usbnet mii mei_me igc thunderbolt udc_core mei pcs_xpcs wdat_wdt irqbypass spi_pxa2xx_platform phylink intel_ish_ipc i2c_i801 dw_dmac tpm_crb intel_rapl_msr pcspkr dw_dmac_core i2c_smbus intel_ishtp tpm_tis parport_pc intel_pmc_core igen6_edac tpm_tis_core thermal parport i8042 tpm edac_core dwc3_pci video drm_buddy ttm drm_display_helper fuse configfs snd_hda_intel snd_intel_dspcfg snd_intel_sdw_acpi snd_hda_codec snd_hda_core snd_pcm snd_timer snd soundcore
> [ 2372.092427] CR2: 0000000000000008
> [ 2372.095744] ---[ end trace 0000000000000000 ]---
> [ 2372.276814] RIP: 0010:taprio_leaf+0x1e/0x30 [sch_taprio]
> [ 2372.282140] Code: Unable to access opcode bytes at RIP 0xffffffffc1112ff4.
> [ 2372.288995] RSP: 0018:ffffae7f82263a28 EFLAGS: 00010206
> [ 2372.294215] RAX: 0000000000000000 RBX: 0000000001000002 RCX: ffff8a704ff38000
> [ 2372.301329] RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff8a7043f29000
> [ 2372.308442] RBP: ffff8a7043f29000 R08: 0000000000000000 R09: ffffae7f82263b98
> [ 2372.315550] R10: 0000000000000002 R11: ffff8a7045df0000 R12: ffff8a70519cc400
> [ 2372.322663] R13: ffffae7f82263b98 R14: ffffffff9a04fb00 R15: ffff8a704ff38000
> [ 2372.329774] FS:  00007fd2fc262740(0000) GS:ffff8a7790a00000(0000) knlGS:0000000000000000
> [ 2372.337840] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [ 2372.343569] CR2: ffffffffc1112ff4 CR3: 00000001053ac003 CR4: 0000000000770ef0
> [ 2372.350680] PKRU: 55555554
> 
> Fixes: 1461d212ab27 ("net/sched: taprio: make qdisc_leaf() see the per-netdev-queue pfifo child qdiscs")
> Reported-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@...el.com>
> Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@...el.com>
> ---
> This patch is more to make others aware of this issue and to explain
> what I think is going on. Perhaps the correct solution is something
> else. (and that explicit refcount seems wrong, but getting warnings
> without it).
> 
> I am questioning my sanity, I have clear memories about testing the
> commit above and it working fine.
> 
> Reverting commit 1461d212ab27 is another alternative, but I think it's
> worth to keep that and fix the problem that it exposed.
> 
> 
>  net/sched/sch_taprio.c | 9 ++-------
>  1 file changed, 2 insertions(+), 7 deletions(-)
> 
> diff --git a/net/sched/sch_taprio.c b/net/sched/sch_taprio.c
> index 86675a79da1e..75aacfff2d5e 100644
> --- a/net/sched/sch_taprio.c
> +++ b/net/sched/sch_taprio.c
> @@ -1763,12 +1763,6 @@ static void taprio_attach(struct Qdisc *sch)
>  		if (old)
>  			qdisc_put(old);
>  	}
> -
> -	/* access to the child qdiscs is not needed in offload mode */
> -	if (FULL_OFFLOAD_IS_ENABLED(q->flags)) {
> -		kfree(q->qdiscs);
> -		q->qdiscs = NULL;
> -	}
>  }

omfg, I didn't include this hunk in the patch? yikes, sorry.

>  
>  static struct netdev_queue *taprio_queue_get(struct Qdisc *sch,
> @@ -1801,8 +1795,9 @@ static int taprio_graft(struct Qdisc *sch, unsigned long cl,
>  		*old = dev_graft_qdisc(dev_queue, new);
>  	} else {
>  		*old = q->qdiscs[cl - 1];
> -		q->qdiscs[cl - 1] = new;
>  	}
> +	q->qdiscs[cl - 1] = new;
> +	qdisc_refcount_inc(new);

What warning are you getting without this part of the change?

>  
>  	if (new)
>  		new->flags |= TCQ_F_ONETXQUEUE | TCQ_F_NOPARENT;
> -- 
> 2.37.3
>

Powered by blists - more mailing lists