NetBSD Problem Report #49536
From o.vd.linden@quicknet.nl Sun Jan 4 19:42:28 2015
Return-Path: <o.vd.linden@quicknet.nl>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
(Client CN "mail.netbsd.org", Issuer "Postmaster NetBSD.org" (verified OK))
by mollari.NetBSD.org (Postfix) with ESMTPS id 4B3C3A5B2E
for <gnats-bugs@gnats.NetBSD.org>; Sun, 4 Jan 2015 19:42:28 +0000 (UTC)
Message-Id: <20150104181857.GA2485@sheep>
Date: Sun, 4 Jan 2015 19:18:57 +0100
From: Onno van der Linden <o.vd.linden@quicknet.nl>
To: gnats-bugs@netbsd.org
Subject: broken drm since changes on dec 14 2014
>Number: 49536
>Category: kern
>Synopsis: radeondrm won't work with sources after 14-12-2014 23:40
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: chs
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Sun Jan 04 19:45:00 +0000 2015
>Closed-Date: Wed Jan 14 05:54:53 +0000 2015
>Last-Modified: Wed Jan 14 05:54:53 +0000 2015
>Originator: Onno van der Linden
>Release: NetBSD 7.99.4
>Organization:
>Environment:
System: NetBSD sheep 7.99.4 NetBSD 7.99.4 (SHEEP) #3: Sun Jan 4 13:58:28 MET 2015 onno@sheep:/usr/src/sys/arch/i386/compile/SHEEP i386
Architecture: i386
Machine: i386
>Description:
Since the drm related mmap changes on Dec 14 2014 the non kms radeondrm driver
won't start X anymore on my i386 with an AGP HD3450.
>How-To-Repeat:
run startx
>Fix:
1) uvm_mmap_dev is called with an uninitialized variable, set it to NULL
--- /sys/external/bsd/drm/dist/bsd-core/drm_bufs.c.orig 2015-01-03 15:01:03.000000000 +0100
+++ /sys/external/bsd/drm/dist/bsd-core/drm_bufs.c 2015-01-04 15:36:34.000000000 +0100
@@ -1122,6 +1122,7 @@
#elif defined(__NetBSD__)
/* XXXNETBSD */
rsize = round_page(size);
+ addr = NULL;
retcode = uvm_mmap_dev(curproc, &addr, rsize, dev->kdev, foff);
vaddr = (vaddr_t)addr;
DRM_DEBUG("mmap %#lx/%#lx foff %#llx\n", vaddr, rsize, (long long)foff);
2) Kernel with DRM_DEBUG and fix above now gets
[drm_mmap] [drm_mmap] Can't find map, requested offset = 0
and
[ 31.716] (II) RADEON(0): [drm] Added 32 65536 byte vertex/indirect buffers
[ 31.716] (EE) RADEON(0): [drm] Failed to map vertex/indirect buffers list
[ 31.725] (II) RADEON(0): [drm] removed 1 reserved context for kernel
[ 31.725] (II) RADEON(0): [drm] unmapping 8192 bytes of SAREA 0xc2987000 at 0xbb400000
[ 31.725] (II) RADEON(0): [drm] Closed DRM master.
[ 31.725] (WW) RADEON(0): Direct rendering disabled
[ 31.733] (EE) RADEON(0): Acceleration initialization failed
[ 31.733] (II) RADEON(0): Acceleration disabled
# diff -u `pwd`/uvm_mmap.c.orig `pwd`/uvm_mmap.c
--- /sys/uvm/uvm_mmap.c.orig 2015-01-03 15:27:57.000000000 +0100
+++ /sys/uvm/uvm_mmap.c 2015-01-04 18:23:12.000000000 +0100
@@ -1084,7 +1084,7 @@
*addrp = (void *)p->p_emul->e_vm_default_addr(p,
(vaddr_t)p->p_vmspace->vm_daddr, len);
- uobj = udv_attach(dev, prot, 0, len);
+ uobj = udv_attach(dev, prot, off, len);
if (uobj == NULL)
return EINVAL;
3) with the 2 previous fixes I now get
[drm:pid201:drm_ioctl] [drm:pid201:drm_ioctl] pid=201, cmd=0xc0286429, nr=0x29, dev 0xc2c9a200, auth=1
[drm:pid201:radeon_freelist_get] [drm:pid201:radeon_freelist_get] done_age = -559038737
[drm:pid201:drm_mmap] [drm:pid201:drm_mmap] Can't find map, requested offset = 10000
[drm:pid201:drm_mmap] [drm:pid201:drm_mmap] map offset = 00000000c2984000, handle = 00000000c2984000
[drm:pid201:drm_mmap] [drm:pid201:drm_mmap] map offset = 00000000df000000, handle = 00000000db700000
[drm:pid201:drm_mmap] [drm:pid201:drm_mmap] map offset = 00000000e0000000, handle = 0000000000000000
[drm:pid201:drm_mmap] [drm:pid201:drm_mmap] map offset = 00000000db710000, handle = 00000000db710000
[drm:pid201:drm_mmap] [drm:pid201:drm_mmap] map offset = 00000000db811000, handle = 00000000db811000
[drm:pid201:drm_mmap] [drm:pid201:drm_mmap] map offset = 00000000db812000, handle = 00000000db812000
[drm:pid201:drm_mmap] [drm:pid201:drm_mmap] map offset = 00000000dba12000, handle = 0000000000000000
That offset is way too small.
And then it all crashes due to a SIGSEGV in r600_start_3d()
(And why is the DRM_DEBUG macro printing stuff twice ?)
>Release-Note:
>Audit-Trail:
Responsible-Changed-From-To: kern-bug-people->chs
Responsible-Changed-By: chs@NetBSD.org
Responsible-Changed-When: Mon, 05 Jan 2015 19:31:56 +0000
Responsible-Changed-Why:
my fault
From: Onno van der Linden <o.vd.linden@quicknet.nl>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/49536
Date: Tue, 6 Jan 2015 23:00:22 +0100
My previous fix for uvm_mmap_dev() needs an
additional fix: the offset parameter in the call
to uvm_mmap(). With the complete fix below I'm
not getting the "Can't find map" drm_debug output.
There's still a "segmentation violation
at address 0xa8xxxxxx" somewhere (at exit ?)
which doesn't crash X. Not sure yet where that
one comes from.
--- uvm_mmap.c.orig 2015-01-03 15:27:57.000000000 +0100
+++ uvm_mmap.c 2015-01-06 22:42:51.000000000 +0100
@@ -1084,13 +1084,13 @@
*addrp = (void *)p->p_emul->e_vm_default_addr(p,
(vaddr_t)p->p_vmspace->vm_daddr, len);
- uobj = udv_attach(dev, prot, 0, len);
+ uobj = udv_attach(dev, prot, off, len);
if (uobj == NULL)
return EINVAL;
error = uvm_mmap(&p->p_vmspace->vm_map, (vaddr_t *)addrp,
(vsize_t)len, prot, prot, flags, UVM_ADV_RANDOM,
- uobj, 0, p->p_rlimit[RLIMIT_MEMLOCK].rlim_cur);
+ uobj, off, p->p_rlimit[RLIMIT_MEMLOCK].rlim_cur);
return error;
}
Onno
From: matthew green <mrg@eterna.com.au>
To: gnats-bugs@NetBSD.org
Cc: chs@NetBSD.org, gnats-admin@netbsd.org, netbsd-bugs@netbsd.org,
Onno van der Linden <o.vd.linden@quicknet.nl>
Subject: re: kern/49536
Date: Wed, 07 Jan 2015 12:28:31 +1100
i can confirm that with this patch (and the prior one for drm_bufs.c)
allows my radeon 4670 laptop to run with old drm again.
From: "Chuck Silvers" <chs@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/49536 CVS commit: src/sys/external/bsd/drm/dist/bsd-core
Date: Sat, 10 Jan 2015 23:34:49 +0000
Module Name: src
Committed By: chs
Date: Sat Jan 10 23:34:49 UTC 2015
Modified Files:
src/sys/external/bsd/drm/dist/bsd-core: drm_bufs.c
Log Message:
initialize "addr" before passing its address to uvm_mmap_dev().
from Onno van der Linden in PR 49536.
To generate a diff of this commit:
cvs rdiff -u -r1.12 -r1.13 src/sys/external/bsd/drm/dist/bsd-core/drm_bufs.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: "Chuck Silvers" <chs@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/49536 CVS commit: src/sys/uvm
Date: Sat, 10 Jan 2015 23:35:02 +0000
Module Name: src
Committed By: chs
Date: Sat Jan 10 23:35:02 UTC 2015
Modified Files:
src/sys/uvm: uvm_mmap.c
Log Message:
in uvm_mmap_dev(), use the passed-in offset instead of 0.
from Onno van der Linden in PR 49536.
To generate a diff of this commit:
cvs rdiff -u -r1.150 -r1.151 src/sys/uvm/uvm_mmap.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: "Soren Jacobsen" <snj@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/49536 CVS commit: [netbsd-7] src/sys/external/bsd/drm/dist/bsd-core
Date: Sun, 11 Jan 2015 06:25:31 +0000
Module Name: src
Committed By: snj
Date: Sun Jan 11 06:25:31 UTC 2015
Modified Files:
src/sys/external/bsd/drm/dist/bsd-core [netbsd-7]: drm_bufs.c
Log Message:
Pull up following revision(s) (requested by chs in ticket #402):
sys/external/bsd/drm/dist/bsd-core/drm_bufs.c: revision 1.13
initialize "addr" before passing its address to uvm_mmap_dev().
from Onno van der Linden in PR 49536.
To generate a diff of this commit:
cvs rdiff -u -r1.11.14.1 -r1.11.14.2 \
src/sys/external/bsd/drm/dist/bsd-core/drm_bufs.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: "Soren Jacobsen" <snj@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/49536 CVS commit: [netbsd-7] src/sys/uvm
Date: Sun, 11 Jan 2015 06:27:40 +0000
Module Name: src
Committed By: snj
Date: Sun Jan 11 06:27:40 UTC 2015
Modified Files:
src/sys/uvm [netbsd-7]: uvm_mmap.c
Log Message:
Pull up following revision(s) (requested by chs in ticket #403):
sys/uvm/uvm_mmap.c: revision 1.151
in uvm_mmap_dev(), use the passed-in offset instead of 0.
from Onno van der Linden in PR 49536.
To generate a diff of this commit:
cvs rdiff -u -r1.148.4.1 -r1.148.4.2 src/sys/uvm/uvm_mmap.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: Chuck Silvers <chuq@chuq.com>
To: gnats-bugs@NetBSD.org
Cc: mrg@netbsd.org
Subject: Re: kern/49536: broken drm since changes on dec 14 2014
Date: Tue, 13 Jan 2015 10:25:49 -0800
the remaining problem with the X server getting a SEGV when exiting
appears to be due to something not getting pulled to the netbsd-7 branch.
this problem doesn't happen with -current userland (even with a
netbsd-7 kernel).
-Chuck
From: Onno van der Linden <o.vd.linden@quicknet.nl>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/49536: broken drm since changes on dec 14 2014
Date: Tue, 13 Jan 2015 20:02:21 +0100
On Tue, Jan 13, 2015 at 06:30:01PM +0000, Chuck Silvers wrote:
> the remaining problem with the X server getting a SEGV when exiting
> appears to be due to something not getting pulled to the netbsd-7 branch.
> this problem doesn't happen with -current userland (even with a
> netbsd-7 kernel).
After recompiling the Xorg server with several older xsrc source trees
I wasn't able to reproduce the SEGV. Then I proceeded to build the Xorg
server with the -current sources: no SEGV either ........
Call can be closed as far as I'm concerned.
Onno
From: "John D. Baker" <jdbaker@mylinuxisp.com>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/49536: broken drm since changes on dec 14 2014
Date: Tue, 13 Jan 2015 16:30:18 -0600 (CST)
Not radeon, but related:
This commit:
http://mail-index.netbsd.org/source-changes/2014/12/14/msg061433.html
Causes X initialization to fail to complete when running a kernel using
the UMS r128drm kernel driver from 14-Dec-2014 23:49 UTC. It then
proceeds to fill up the "/var" filesystem with Xorg.0.log ending in a
series of messages like:
[...]
[ 123.364] (II) AIGLX: Screen 0 is not DRI2 capable
[ 123.364] (II) AIGLX: Screen 0 is not DRI capable
[ 123.656] (II) AIGLX: Loaded and initialized /usr/X11R7/lib/modules/dri/swrast_dri.so
[ 123.656] (II) GLX: Initialized DRISWRAST GL provider for screen 0
[ 123.813] (EE) R128(0): R128CCEWaitForIdle: CCE idle -9
[ 123.813] (EE) R128(0): Idle timed out, resetting engine...
[ 123.813] (EE) R128(0): R128CCEWaitForIdle: CCE stop -9
[ 123.813] (EE) R128(0): R128CCEWaitForIdle: CCE start -9
[ 123.813] (EE) R128(0): R128CCEWaitForIdle: CCE idle -9
[ 123.813] (EE) R128(0): Idle timed out, resetting engine...
[ 123.813] (EE) R128(0): R128CCEWaitForIdle: CCE stop -9
[ 123.813] (EE) R128(0): R128CCEWaitForIdle: CCE start -9
[..., repeats until "/var" is out of space, ...]
A UMS r128drm kernel from 14-Dec-2014 23:48 UTC produces a working X
display.
The symptoms persist in the very latest -current and in netbsd-7 following
the pullup of the changes in the above-mentioned commit.
--
|/"\ John D. Baker, KN5UKS NetBSD Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com OpenBSD FreeBSD
| X No HTML/proprietary data in email. BSD just sits there and works!
|/ \ GPGkeyID: D703 4A7E 479F 63F8 D3F4 BD99 9572 8F23 E4AD 1645
--
|/"\ John D. Baker, KN5UKS NetBSD Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com OpenBSD FreeBSD
| X No HTML/proprietary data in email. BSD just sits there and works!
|/ \ GPGkeyID: D703 4A7E 479F 63F8 D3F4 BD99 9572 8F23 E4AD 1645
From: "John D. Baker" <jdbaker@mylinuxisp.com>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/49536: broken drm since changes on dec 14 2014
Date: Tue, 13 Jan 2015 19:54:04 -0600 (CST)
On Tue, 13 Jan 2015, John D. Baker wrote:
> Not radeon, but related:
>
> This commit:
>
> http://mail-index.netbsd.org/source-changes/2014/12/14/msg061433.html
The changes already committed and pulled up for this PR fix the issue
I was seeing as well. (I was still bisecting code, but was unable to
visit the test system's console for almost a week.)
I confirmed the fix on netbsd-7, will do so with -current soon.
Thanks.
--
|/"\ John D. Baker, KN5UKS NetBSD Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com OpenBSD FreeBSD
| X No HTML/proprietary data in email. BSD just sits there and works!
|/ \ GPGkeyID: D703 4A7E 479F 63F8 D3F4 BD99 9572 8F23 E4AD 1645
From: "John D. Baker" <jdbaker@mylinuxisp.com>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/49536: broken drm since changes on dec 14 2014
Date: Tue, 13 Jan 2015 22:21:30 -0600 (CST)
On Tue, 13 Jan 2015, John D. Baker wrote:
> I confirmed the fix on netbsd-7, will do so with -current soon.
Confirmed working.
Thanks.
--
|/"\ John D. Baker, KN5UKS NetBSD Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com OpenBSD FreeBSD
| X No HTML/proprietary data in email. BSD just sits there and works!
|/ \ GPGkeyID: D703 4A7E 479F 63F8 D3F4 BD99 9572 8F23 E4AD 1645
State-Changed-From-To: open->closed
State-Changed-By: msaitoh@NetBSD.org
State-Changed-When: Wed, 14 Jan 2015 05:54:53 +0000
State-Changed-Why:
Fixed and pulled up.
Thanks.
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2014
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.
Home