RE: [EXTERNAL] Re: Kernel patch in USB 4.4.138 Version

[Felipe Balbi]

Hi,

"Varadarajan, Kiruthika" <Kiruthika.Varadarajan@xxxxxxxxxx> writes:
> Hi Greg,
>  Please ignore my previous email and here is the patch on top of 4.4.x stable version.
>  Could you please review the attached patch.
>
> Thanks and Regards,
> Kiruthika. NV
>
> -----Original Message-----
> From: Varadarajan, Kiruthika 
> Sent: Tuesday, June 18, 2019 2:07 PM
> To: 'Greg Kroah-Hartman' <gregkh@xxxxxxxxxxxxxxxxxxx>
> Cc: Felipe Balbi <balbi@xxxxxx>; linux-usb@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
> Subject: RE: [EXTERNAL] Re: Kernel patch in USB 4.4.138 Version
>
>  Hi Greg,
>  I applied the attached patch against 4.4.x kernel version. Could you please review the patch.
>
>
> Thanks and Regards,
> Kiruthika. NV
>
> -----Original Message-----
> From: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> Sent: Friday, June 7, 2019 7:29 PM
> To: Varadarajan, Kiruthika <Kiruthika.Varadarajan@xxxxxxxxxx>
> Cc: Felipe Balbi <balbi@xxxxxx>; linux-usb@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
> Subject: [EXTERNAL] Re: Kernel patch in USB 4.4.138 Version
>
> On Fri, Jun 07, 2019 at 08:23:37AM +0000, Varadarajan, Kiruthika wrote:
>> Hi All,
>>   During Cay Play session connect/disconnect several times,  with USB gadget driver I found kernel crash in rx_submit.
>> 
>> 
>> 64.802125] CPU: 0 PID: 636 Comm: kworker/u8:2 Tainted: G U W O 
>> 4.4.71-quilt-2e5dc0ac-g0b463804d692-dirty #45 [ 64.814092] Workqueue:
>> uether process_rx_w [ 64.818675] task: ffff8800779c40c0 ti: 
>> ffff88007702c000 task.ti: ffff88007702c000 [ 64.827040] RIP: 
>> 0010:[<ffffffff81736b6d>] [<ffffffff81736b6d>] rx_fill+0x1ed/0x3b0 [ 
>> 64.835706] RSP: 0000:ffff88007702fd80 EFLAGS: 00010246 [ 64.841642]
>> RAX: 0000000000000000 RBX: ffff88006f2768c0 RCX: 0000000000000200 [ 
>> 64.849617] RDX: 00000000000001fd RSI: ffffffff82081f71 RDI:
>> 00000000ffffffff [ 64.857591] RBP: ffff88007702fdc8 R08: 
>> ffffc90000285000 R09: 0000000000000600 [ 64.865566] R10: 
>> ffff88002ab99670 R11: 0000000000000f7a R12: 0000000000000002 [ 
>> 64.873541] R13: ffff88006f2768f0 R14: 0000000000000600 R15:
>> ffff88006bf6f118 [ 64.881509] FS: 0000000000000000(0000)
>> GS:ffff88007ac00000(0000) knlGS:0000000000000000 [ 64.890545] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 64.896965] CR2: 00000000000000f0 CR3: 0000000065c99000 CR4: 00000000003406f0 [ 64.904936] Last Branch Records:
>> [ 64.908551] to: [<ffffffff81a50260>] page_fault+0x0/0x80 [ 64.914787]
>> from: [<ffffffff81736b6d>] rx_fill+0x1ed/0x3b0 [ 64.921020] to: 
>> [<ffffffff81736b69>] rx_fill+0x1e9/0x3b0 [ 64.927253] from: 
>> [<ffffffff81736b63>] rx_fill+0x1e3/0x3b0 [ 64.933480] to: 
>> [<ffffffff81736b37>] rx_fill+0x1b7/0x3b0 [ 64.939704] from: 
>> [<ffffffff811804bd>] printk+0x4f/0x50 [ 64.945654] to: 
>> [<ffffffff811804b6>] printk+0x48/0x50 [ 64.951588] from: 
>> [<ffffffff810ecb60>] vprintk_default+0x20/0x30 [ 64.958403] to: 
>> [<ffffffff810ecb5f>] vprintk_default+0x1f/0x30 [ 64.965219] from: 
>> [<ffffffff810ec64f>] vprintk_emit+0x16f/0x530 [ 64.971941] to: 
>> [<ffffffff810ec641>] vprintk_emit+0x161/0x530 [ 64.978653] from: 
>> [<ffffffff810ec752>] vprintk_emit+0x272/0x530 [ 64.985383] to: 
>> [<ffffffff810ec752>] vprintk_emit+0x272/0x530 [ 64.992116] from: 
>> [<ffffffff81001387>] restore+0xd/0x16 [ 64.998059] to: 
>> [<ffffffff8100137a>] restore+0x0/0x16 [ 65.003999] from: 
>> [<ffffffff81001364>] ___preempt_schedule+0x12/0x14 [ 65.011205] to: 
>> [<ffffffff81001364>] ___preempt_schedule+0x12/0x14 [ 65.018411] from: 
>> [<ffffffff81a47a60>] preempt_schedule+0x20/0x30 [ 65.025324] to: 
>> [<ffffffff81a47a5f>] preempt_schedule+0x1f/0x30 [ 65.032228] from: 
>> [<ffffffff81a47a33>] preempt_schedule_common+0x33/0x40 [ 65.039835]
>> to: [<ffffffff81a47a1f>] preempt_schedule_common+0x1f/0x40 [ 
>> 65.047427] from: [<ffffffff81a471db>] __schedule+0x36b/0xb90 [ 
>> 65.053952] to: [<ffffffff81a471bc>] __schedule+0x34c/0xb90 [ 
>> 65.060478] from: [<ffffffff810c1326>] finish_task_switch+0xb6/0x210 [ 
>> 65.067688] to: [<ffffffff810c12fb>] finish_task_switch+0x8b/0x210 [ 
>> 65.074886] from: [<ffffffff81a4db3a>] _raw_spin_unlock_irq+0x2a/0x40 [ 
>> 65.082190] to: [<ffffffff81a4db2e>] _raw_spin_unlock_irq+0x1e/0x40 [ 
>> 65.089494] from: [<ffffffff810bde65>] preempt_count_sub+0x35/0xb0 [ 
>> 65.096508] to: [<ffffffff810bde5b>] preempt_count_sub+0x2b/0xb0 [ 
>> 65.103511] from: [<ffffffff810bdea5>] preempt_count_sub+0x75/0xb0 [ 
>> 65.110517] to: [<ffffffff810bde9c>] preempt_count_sub+0x6c/0xb0 [ 
>> 65.117522] from: [<ffffffff810bde59>] preempt_count_sub+0x29/0xb0 [ 
>> 65.124530] to: [<ffffffff810bde30>] preempt_count_sub+0x0/0xb0 [ 
>> 65.131444] from: [<ffffffff81a4db29>] _raw_spin_unlock_irq+0x19/0x40 [ 
>> 65.138748] to: [<ffffffff81a4db23>] _raw_spin_unlock_irq+0x13/0x40 [ 
>> 65.146052] from: [<ffffffff81155389>] trace_hardirqs_on+0x9/0xf0 [ 
>> 65.152958] to: [<ffffffff81155380>] trace_hardirqs_on+0x0/0xf0 [ 
>> 65.159872] from: [<ffffffff81a4db1e>] _raw_spin_unlock_irq+0xe/0x40 [ 
>> 65.167065] to: [<ffffffff81a4db1e>] _raw_spin_unlock_irq+0xe/0x40 [ 
>> 65.174257] from: [<ffffffff810e58cb>] do_raw_spin_unlock+0x4b/0xa0 [ 
>> 65.181364] to: [<ffffffff810e5880>] do_raw_spin_unlock+0x0/0xa0 [ 
>> 65.188372] from: [<ffffffff81a4db19>] _raw_spin_unlock_irq+0x9/0x40 [ 
>> 65.195575] to: [<ffffffff81a4db10>] _raw_spin_unlock_irq+0x0/0x40 [ 
>> 65.202776] from: [<ffffffff810c12f6>] finish_task_switch+0x86/0x210 [ 
>> 65.209980] to: [<ffffffff810c1295>] finish_task_switch+0x25/0x210 [ 
>> 65.217183] from: [<ffffffff813dec08>] debug_smp_processor_id+0x18/0x20 
>> [ 65.224676] to: [<ffffffff813dec07>] debug_smp_processor_id+0x17/0x20 
>> [ 65.232171] from: [<ffffffff813deb26>]
>> check_preemption_disabled+0x26/0xf0
>> [ 65.239952] to: [<ffffffff813deb00>]
>> check_preemption_disabled+0x0/0xf0
>> [ 65.247632] from: [<ffffffff813dec02>]
>> debug_smp_processor_id+0x12/0x20 [ 65.255124] to: [<ffffffff813debf0>]
>> debug_smp_processor_id+0x0/0x20 [ 65.262521] from: 
>> [<ffffffff810c1290>] finish_task_switch+0x20/0x210 [ 65.269734] to: 
>> [<ffffffff810c1270>] finish_task_switch+0x0/0x210 [ 65.276831] from: 
>> [<ffffffff81a471b7>] __schedule+0x347/0xb90 [ 65.283353] to: 
>> [<ffffffff81a47184>] __schedule+0x314/0xb90 [ 65.289871] from: 
>> [<ffffffff81003677>] __switch_to+0x297/0x5a0 [ 65.296486] to: 
>> [<ffffffff81003666>] __switch_to+0x286/0x5a0 [ 65.303092] from: 
>> [<ffffffff81003654>] __switch_to+0x274/0x5a0 [ 65.309728] to: 
>> [<ffffffff81003610>] __switch_to+0x230/0x5a0 [ 65.316352] from: 
>> [<ffffffff81003600>] __switch_to+0x220/0x5a0 [ 65.322972] to: 
>> [<ffffffff810035d3>] __switch_to+0x1f3/0x5a0 [ 65.329582] from: 
>> [<ffffffff810035c3>] __switch_to+0x1e3/0x5a0 [ 65.336203] to: 
>> [<ffffffff81003525>] __switch_to+0x145/0x5a0 [ 65.342821] from: 
>> [<ffffffff810034ce>] __switch_to+0xee/0x5a0 [ 65.349343] Stack:
>> [ 65.351585] 0000000000000286 ffff88006bd22548 024000c081a4dae8
>> ffff88006f276918 [ 65.359856] 0000000000000000 0000000000000000
>> ffff88006f276938 ffff88006f2768c0 [ 65.368143] ffff88006f2769a8
>> ffff88007702fe00 ffffffff81736f5f ffff88006f2769a8 [ 65.376447] Call Trace:
>> [ 65.379176] [<ffffffff81736f5f>] process_rx_w+0xcf/0x130 [ 65.385213] 
>> [<ffffffff810b29aa>] process_one_work+0x14a/0x450 [ 65.391736] 
>> [<ffffffff810b2fe6>] worker_thread+0x66/0x450 [ 65.397868] 
>> [<ffffffff810b2f80>] ? rescuer_thread+0x2d0/0x2d0 [ 65.404391] 
>> [<ffffffff810b2f80>] ? rescuer_thread+0x2d0/0x2d0 [ 65.410911] 
>> [<ffffffff810b80ad>] kthread+0xfd/0x120 [ 65.416456] 
>> [<ffffffff810b7fb0>] ? kthread_create_on_node+0x190/0x190
>> [ 65.423755] [<ffffffff81a4e46f>] ret_from_fork+0x3f/0x70 [ 65.429785] 
>> [<ffffffff810b7fb0>] ? kthread_create_on_node+0x190/0x190
>> [ 65.437077] Code: 4c 89 c9 41 83 e9 01 4d 63 c9 4d 01 f1 4c 89 c8 48
>> f7 f1 8b 83 b0 00 00 00 49 29 d1 4d 89 ce 85 c0 74 04 4c 0f af f0 48 
>> 8b 43 18 <80> b8 f0 00 00 00 00 0f 85 78 01 00 00 48 8b 73 20 4c 89 f1
>> 48 [ 65.458462] RIP [<ffffffff81736b6d>] rx_fill+0x1ed/0x3b0 [ 
>> 65.464493] RSP <ffff88007702fd80> [ 65.468388] CR2: 00000000000000f0 [ 
>> 65.472090] ---[ end trace 5a9d1e73247ee2aa ]--- [ 65.479952] Kernel 
>> panic - not syncing: Fatal exception [ 65.486167] intel_tasks_dumper: intel_task_panic_hndl: Skip tasks dumper.
>> [ 65.493763] reboot: panic mode set: p,w [ 65.498045] Kernel Offset: 
>> disabled [ 65.509960] Rebooting in 10 seconds..
>> 
>> 
>>   I have a patch and I would like to mainline the same patch.
>>   Could you please review the same and provide your comments.
>
> Please submit this in a format that we can apply it in.  We can't apply patchs directly to the 4.4.y tree, they have to be backports from Linus's tree only (read the stable kernel rules for details.)
>
> Also, your patch referenced a git commit id that is not in Linus's tree, and you do not need the Change-Id: line as well.
>
> Please fix that up, and redo it against linux-next and resend and we will be glad to review it.
>
> thanks,
>
> greg k-h
> From b45595c42b75845bac54e11b32d93f396b8c0745 Mon Sep 17 00:00:00 2001
> From: kvaradarajan <Kiruthika.Varadarajan@xxxxxxxxxx>
> Date: Tue, 18 Jun 2019 08:39:06 +0000
> Subject: [PATCH]  USB: Fix race between gether_disconnect and rx_submit
>
>   On spin lock release in rx_submit, gether_disconnect get
>   a chance to run, it makes port_usb NULL, rx_submit access
>   NULL port USB, hence null pointer crash.
>   Fixed by releasing the lock in rx_submit after port_usb
>   is used.
>
>   Signed-off-by:Kiruthika Varadarajan<Kiruthika.Varadarajan@xxxxxxxxxx>
> ---
>  drivers/usb/gadget/function/u_ether.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/usb/gadget/function/u_ether.c b/drivers/usb/gadget/function/u_ether.c
> index 7413f89..e69f20b 100644
> --- a/drivers/usb/gadget/function/u_ether.c
> +++ b/drivers/usb/gadget/function/u_ether.c
> @@ -207,11 +207,12 @@ rx_submit(struct eth_dev *dev, struct usb_request *req, gfp_t gfp_flags)
>  		out = dev->port_usb->out_ep;
>  	else
>  		out = NULL;
> -	spin_unlock_irqrestore(&dev->lock, flags);
>  
>  	if (!out)
> +	{
> +		spin_unlock_irqrestore(&dev->lock, flags);
>  		return -ENOTCONN;
> -
> +	}
>  
>  	/* Padding up to RX_EXTRA handles minor disagreements with host.
>  	 * Normally we use the USB "terminate on short read" convention;
> @@ -232,6 +233,7 @@ rx_submit(struct eth_dev *dev, struct usb_request *req, gfp_t gfp_flags)
>  
>  	if (dev->port_usb->is_fixed)
>  		size = max_t(size_t, size, dev->port_usb->fixed_out_len);
> +	spin_unlock_irqrestore(&dev->lock, flags);
>  
>  	skb = alloc_skb(size + NET_IP_ALIGN, gfp_flags);
>  	if (skb == NULL) {

Thanks, this is a good fix. dev->port_usb should, indeed, be accessed
with spin lock held. I have taken your patch from the attachment this
time around, but keep in mind that this is not how we send patches. We
have documentation explaining how to do so. Also, there were some issues
with your commit log, subject line and signed-off-by line. I've fixed
them all.

I'll resend the fixed up version of the patch so you see how it
should've been.

cheers

-- 
balbi

Attachment: signature.asc
Description: PGP signature