kern/69319: aue(4) startup panic

[Ian Dowse]

The following reply was made to PR kern/69319; it has been noted by GNATS.

From: Ian Dowse <iedowse at maths.tcd.ie>
To: Scott Mitchell <scott+freebsd at fishballoon.org>
Cc: julian at freebsd.org, FreeBSD-gnats-submit at freebsd.org
Subject: Re: kern/69319: aue(4) startup panic 
Date: Tue, 20 Jul 2004 01:32:35 +0100

 In message <200407192324.i6JNOOQt000616 at tuatara.fishballoon.org>, Scott Mitchel
 l writes:
 > db> tr
 > tsleep(c20b3f00,10,c0431b18,0) at tsleep+0x74
 > usbd_transfer(c20b3f00,c044fff4,c036f10d,c20b3f00,0) at usbd_transfer+0x127
 > usbd_sync_transfer(c20b3f00,0,c1df5200,1,c03b784c) at usbd_sync_transfer+0x10
 > usbd_do_request_flags_pipe(c1ecc000,c1ed7f80,c045004c,c0450064,0) at usbd_do_
 >request_flags_pipe+0x61
 > usbd_do_request_flags(c1ecc000,c045004c,c0450064,0,0) at usbd_do_request_flag
 >s+0x20
 > usbd_do_request(c1ecc000,c045004c,c0450064,f140,10025) at usbd_do_request+0x1
 >a
 > aue_csr_write_1(c1df5200,25,0,1,0) at aue_csr_write_1+0x6c
 
 It looks as if the USBD_NO_TSLEEP hack was accidentally removed
 during the big USB MFC. I believe that this is still needed in
 RELENG_4.
 
 Bringing it back may just involves undoing the small bits of the
 following revisions that mention that flag, though there might be
 more to it:
 
 	usbdi.h 1.21.2.4
 	usbdi.c 1.34.2.8
 	if_rue.c 1.2.2.2
 	if_aue.c 1.19.2.20
 	if_cue.c 1.7.2.7
 	if_kue.c 1.17.2.10
 
 It may be easier to reverse-apply the corresponding revisions from
 -CURRENT and fix the conflicts, i.e.:
 
 	cvs update -kk -j1.28 -j1.27 if_cue.c
 	cvs update -kk -j1.56 -j1.55 if_aue.c
 	cvs update -kk -j1.40 -j1.39 if_kue.c
 	cvs update -kk -j1.65 -j1.64 usbdi.c
 	cvs update -kk -j1.41 -j1.40 usbdi.h
 	(there is no suitable revision for if_rue.c)
 
 Ian