xhci: Cancel cmd_timer before cleanup the command queue in xhci_hc_died()
Subsystems (2)
PATCH 1/1 xhci: Cancel cmd_timer before cleanup the command queue in xhci_hc_died()
When the cmd_timer fired, it would try to access the command struct. So cancel it before cleanup the command queue in xhci_hc_died(), to avoid use-after-free reported by KASAN:
[ 176.952537] BUG: KASAN: use-after-free in xhci_handle_command_timeout+0x68/0x224 [ 176.960846] Write of size 4 at addr ffffffc0cbb01608 by task kworker/3:3/1680 ... [ 177.180644] Freed by task 0: [ 177.183882] kasan_slab_free+0x90/0x15c [ 177.188194] kfree+0x114/0x28c [ 177.191630] xhci_cleanup_command_queue+0xc8/0xf8 [ 177.196916] xhci_hc_died+0x84/0x358
---
drivers/usb/host/xhci-ring.c | 1 + 1 file changed, 1 insertion(+)
drivers/usb/host/xhci-ring.c +1 -0
@@ -914,6 +914,7 @@ void xhci_hc_died(struct xhci_hcd *xhci)xhci_err (xhci,"xHCI host controller not responding, assume dead\n" ); xhci->xhc_state |=XHCI_STATE_DYING ; +cancel_delayed_work_sync (&xhci->cmd_timer);xhci_cleanup_command_queue (xhci);/* return any pending urbs, remove may be waiting for them */
Signature · 1 line
2.11.0
Greg Kroah-Hartman
maintainerQuoted 12 lines · Jeffy Chen
When the
cmd_timerfired, it would try to access the command struct. So cancel it before cleanup the command queue inxhci_hc_died(), to avoid use-after-free reported by KASAN:[ 176.952537] BUG: KASAN: use-after-free in xhci_handle_command_timeout+0x68/0x224 [ 176.960846] Write of size 4 at addr ffffffc0cbb01608 by task kworker/3:3/1680 ... [ 177.180644] Freed by task 0: [ 177.183882] kasan_slab_free+0x90/0x15c [ 177.188194] kfree+0x114/0x28c [ 177.191630] xhci_cleanup_command_queue+0xc8/0xf8 [ 177.196916] xhci_hc_died+0x84/0x358
Quoted 3 lines
Fixes d9f11ba9f107 ("xhci: Rework how we handle unresponsive or hoptlug removed hosts")Signed-off-by Jeffy Chen <[email protected]>---
Mathias, did you miss this one? Want me to queue it up?
thanks,
greg k-h
Mathias Nyman
maintainerQuoted 13 lines · Greg Kroah-Hartman
Quoted 12 lines · Jeffy Chen
When the
cmd_timerfired, it would try to access the command struct. So cancel it before cleanup the command queue inxhci_hc_died(), to avoid use-after-free reported by KASAN:[ 176.952537] BUG: KASAN: use-after-free in xhci_handle_command_timeout+0x68/0x224 [ 176.960846] Write of size 4 at addr ffffffc0cbb01608 by task kworker/3:3/1680 ... [ 177.180644] Freed by task 0: [ 177.183882] kasan_slab_free+0x90/0x15c [ 177.188194] kfree+0x114/0x28c [ 177.191630] xhci_cleanup_command_queue+0xc8/0xf8 [ 177.196916] xhci_hc_died+0x84/0x358
Quoted 6 lines
Quoted 3 lines
Fixes d9f11ba9f107 ("xhci: Rework how we handle unresponsive or hoptlug removed hosts")Signed-off-by Jeffy Chen <[email protected]>---
Mathias, did you miss this one? Want me to queue it up?
There is a v2 of that patch already upstreamed.
d1aad52 xhci: Cleanup current_cmd in xhci_cleanup_command_queue()
https://marc.info/?l=linux-usb&m=150668331802206&w=2
A bit confusing as patch subject changed as well in v2
-Mathias
Greg Kroah-Hartman
maintainerQuoted 21 lines · Mathias Nyman
Quoted 13 lines · Greg Kroah-Hartman
Quoted 12 lines · Jeffy Chen
When the
cmd_timerfired, it would try to access the command struct. So cancel it before cleanup the command queue inxhci_hc_died(), to avoid use-after-free reported by KASAN:[ 176.952537] BUG: KASAN: use-after-free in xhci_handle_command_timeout+0x68/0x224 [ 176.960846] Write of size 4 at addr ffffffc0cbb01608 by task kworker/3:3/1680 ... [ 177.180644] Freed by task 0: [ 177.183882] kasan_slab_free+0x90/0x15c [ 177.188194] kfree+0x114/0x28c [ 177.191630] xhci_cleanup_command_queue+0xc8/0xf8 [ 177.196916] xhci_hc_died+0x84/0x358Quoted 6 lines
Quoted 3 lines
Fixes d9f11ba9f107 ("xhci: Rework how we handle unresponsive or hoptlug removed hosts")Signed-off-by Jeffy Chen <[email protected]>---
Mathias, did you miss this one? Want me to queue it up?
Quoted 7 lines
There is a v2 of that patch already upstreamed.
d1aad52 xhci: Cleanup
current_cmdinxhci_cleanup_command_queue()https://marc.info/?l=linux-usb&m=150668331802206&w=2
A bit confusing as patch subject changed as well in v2
Ah, thanks, missed that.
greg k-h