Re: kern/163091: commit references a PR
The following reply was made to PR kern/163091; it has been noted by GNATS. From: dfil...@freebsd.org (dfilter service) To: bug-follo...@freebsd.org Cc: Subject: Re: kern/163091: commit references a PR Date: Wed, 21 Dec 2011 08:46:17 + (UTC) Author: hselasky Date: Wed Dec 21 08:46:08 2011 New Revision: 228758 URL: http://svn.freebsd.org/changeset/base/228758 Log: Fix for race against user-space applications trying to change the configuration on USB HUBs. PR: kern/163091 MFC after: 1 week Modified: head/sys/dev/usb/usb_hub.c Modified: head/sys/dev/usb/usb_hub.c == --- head/sys/dev/usb/usb_hub.c Wed Dec 21 02:45:51 2011(r228757) +++ head/sys/dev/usb/usb_hub.c Wed Dec 21 08:46:08 2011(r228758) @@ -707,6 +707,13 @@ uhub_explore(struct usb_device *udev) DPRINTF(Device is suspended!\n); return (0); } + + /* + * Make sure we don't race against user-space applications + * like LibUSB: + */ + usbd_enum_lock(udev); + for (x = 0; x != hub-nports; x++) { up = hub-ports + x; portno = x + 1; @@ -784,6 +791,8 @@ uhub_explore(struct usb_device *udev) up-restartcnt = 0; } + usbd_enum_unlock(udev); + /* initial status checked */ sc-sc_flags |= UHUB_FLAG_DID_EXPLORE; ___ svn-src-...@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-all To unsubscribe, send any mail to svn-src-all-unsubscr...@freebsd.org ___ freebsd-usb@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-usb To unsubscribe, send any mail to freebsd-usb-unsubscr...@freebsd.org
Re: a few usb issues related to edge cases
on 20/12/2011 14:25 Andriy Gapon said the following: I just wanted to draw your attention to the fact that obtaining any locks in the kdb context (or USB polling code in general, even) is not a good idea. Chances of getting into trouble on those locks are probably quite moderate or even low, but they do exist. I am not sure if you are getting any bug reports about such troubles :-) Regular users probably do not use kdb too often and a panic for them is just a crash, so they likely do not expect anything usable/debuggable after that :-) Looking some more at the code I just got myself confused as to how the dumping to a umass device could work when the scheduler is stopped. It seems that the umass_command_start - usbd_transfer_start - usbd_callback_ss_done_defer functions would always put a transfer request onto a queue and try to wake up a thread to process that queue and the request. But that's obviously not going to work when the other thread is not going to be run. Have I missed a code path that leads directly to the controller in this context? Thank you for your help. -- Andriy Gapon ___ freebsd-usb@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-usb To unsubscribe, send any mail to freebsd-usb-unsubscr...@freebsd.org
Re: a few usb issues related to edge cases
On Wednesday 21 December 2011 12:29:49 Andriy Gapon wrote: on 20/12/2011 14:25 Andriy Gapon said the following: I just wanted to draw your attention to the fact that obtaining any locks in the kdb context (or USB polling code in general, even) is not a good idea. Chances of getting into trouble on those locks are probably quite moderate or even low, but they do exist. I am not sure if you are getting any bug reports about such troubles :-) Regular users probably do not use kdb too often and a panic for them is just a crash, so they likely do not expect anything usable/debuggable after that :-) Looking some more at the code I just got myself confused as to how the dumping to a umass device could work when the scheduler is stopped. It seems that the umass_command_start - usbd_transfer_start - usbd_callback_ss_done_defer functions would always put a transfer request onto a queue and try to wake up a thread to process that queue and the request. But that's obviously not going to work when the other thread is not going to be run. Have I missed a code path that leads directly to the controller in this context? Thank you for your help. Hi, Those threads should be polled when calling usbd_transfer_poll(). I.E. the wakeup should be stubbed in the !scheduler_running case. --HPS ___ freebsd-usb@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-usb To unsubscribe, send any mail to freebsd-usb-unsubscr...@freebsd.org
Re: a few usb issues related to edge cases
on 21/12/2011 18:38 Hans Petter Selasky said the following: On Wednesday 21 December 2011 12:29:49 Andriy Gapon wrote: on 20/12/2011 14:25 Andriy Gapon said the following: I just wanted to draw your attention to the fact that obtaining any locks in the kdb context (or USB polling code in general, even) is not a good idea. Chances of getting into trouble on those locks are probably quite moderate or even low, but they do exist. I am not sure if you are getting any bug reports about such troubles :-) Regular users probably do not use kdb too often and a panic for them is just a crash, so they likely do not expect anything usable/debuggable after that :-) Looking some more at the code I just got myself confused as to how the dumping to a umass device could work when the scheduler is stopped. It seems that the umass_command_start - usbd_transfer_start - usbd_callback_ss_done_defer functions would always put a transfer request onto a queue and try to wake up a thread to process that queue and the request. But that's obviously not going to work when the other thread is not going to be run. Have I missed a code path that leads directly to the controller in this context? Thank you for your help. Hi, Those threads should be polled when calling usbd_transfer_poll(). I.E. the wakeup should be stubbed in the !scheduler_running case. Ah, that's what I missed! -- Andriy Gapon ___ freebsd-usb@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-usb To unsubscribe, send any mail to freebsd-usb-unsubscr...@freebsd.org