USB: yurex: Fix bad gfp argument
authorAlan Stern <stern@rowland.harvard.edu>
Mon, 10 Aug 2020 18:29:54 +0000 (14:29 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 3 Sep 2020 09:21:22 +0000 (11:21 +0200)
commit f176ede3a3bde5b398a6777a7f9ff091baa2d3ff upstream.

The syzbot fuzzer identified a bug in the yurex driver: It passes
GFP_KERNEL as a memory-allocation flag to usb_submit_urb() at a time
when its state is TASK_INTERRUPTIBLE, not TASK_RUNNING:

do not call blocking ops when !TASK_RUNNING; state=1 set at [<00000000370c7c68>] prepare_to_wait+0xb1/0x2a0 kernel/sched/wait.c:247
WARNING: CPU: 1 PID: 340 at kernel/sched/core.c:7253 __might_sleep+0x135/0x190
kernel/sched/core.c:7253
Kernel panic - not syncing: panic_on_warn set ...
CPU: 1 PID: 340 Comm: syz-executor677 Not tainted 5.8.0-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google
01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0xf6/0x16e lib/dump_stack.c:118
 panic+0x2aa/0x6e1 kernel/panic.c:231
 __warn.cold+0x20/0x50 kernel/panic.c:600
 report_bug+0x1bd/0x210 lib/bug.c:198
 handle_bug+0x41/0x80 arch/x86/kernel/traps.c:234
 exc_invalid_op+0x14/0x40 arch/x86/kernel/traps.c:254
 asm_exc_invalid_op+0x12/0x20 arch/x86/include/asm/idtentry.h:536
RIP: 0010:__might_sleep+0x135/0x190 kernel/sched/core.c:7253
Code: 65 48 8b 1c 25 40 ef 01 00 48 8d 7b 10 48 89 fe 48 c1 ee 03 80 3c 06 00 75
2b 48 8b 73 10 48 c7 c7 e0 9e 06 86 e8 ed 12 f6 ff <0f> 0b e9 46 ff ff ff e8 1f
b2 4b 00 e9 29 ff ff ff e8 15 b2 4b 00
RSP: 0018:ffff8881cdb77a28 EFLAGS: 00010282
RAX: 0000000000000000 RBX: ffff8881c6458000 RCX: 0000000000000000
RDX: ffff8881c6458000 RSI: ffffffff8129ec93 RDI: ffffed1039b6ef37
RBP: ffffffff86fdade2 R08: 0000000000000001 R09: ffff8881db32f54f
R10: 0000000000000000 R11: 0000000030343354 R12: 00000000000001f2
R13: 0000000000000000 R14: 0000000000000068 R15: ffffffff83c1b1aa
 slab_pre_alloc_hook.constprop.0+0xea/0x200 mm/slab.h:498
 slab_alloc_node mm/slub.c:2816 [inline]
 slab_alloc mm/slub.c:2900 [inline]
 kmem_cache_alloc_trace+0x46/0x220 mm/slub.c:2917
 kmalloc include/linux/slab.h:554 [inline]
 dummy_urb_enqueue+0x7a/0x880 drivers/usb/gadget/udc/dummy_hcd.c:1251
 usb_hcd_submit_urb+0x2b2/0x22d0 drivers/usb/core/hcd.c:1547
 usb_submit_urb+0xb4e/0x13e0 drivers/usb/core/urb.c:570
 yurex_write+0x3ea/0x820 drivers/usb/misc/yurex.c:495

This patch changes the call to use GFP_ATOMIC instead of GFP_KERNEL.

Reported-and-tested-by: syzbot+c2c3302f9c601a4b1be2@syzkaller.appspotmail.com
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20200810182954.GB307778@rowland.harvard.edu
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/usb/misc/yurex.c

index 2350502f90540195c8e81a5f89ef9bbe8dbaea48..1be1fa1b737703a43289f44fac27d017cf925ba3 100644 (file)
@@ -502,7 +502,7 @@ static ssize_t yurex_write(struct file *file, const char __user *user_buffer,
        prepare_to_wait(&dev->waitq, &wait, TASK_INTERRUPTIBLE);
        dev_dbg(&dev->interface->dev, "%s - submit %c\n", __func__,
                dev->cntl_buffer[0]);
-       retval = usb_submit_urb(dev->cntl_urb, GFP_KERNEL);
+       retval = usb_submit_urb(dev->cntl_urb, GFP_ATOMIC);
        if (retval >= 0)
                timeout = schedule_timeout(YUREX_WRITE_TIMEOUT);
        finish_wait(&dev->waitq, &wait);