Linux 2.5.18-dj1

Linux 2.5.18-dj1

[Dave Jones]

Resync against 2.5.18, and go through the pending patches folder.
Mostly compilation fixes, but a handful of useful bits too.
A lot of the pending stuff doesn't apply any more, so it may
take a while for me to get around to applying. (ie, resending
resynced versions of patches you sent me may be quicker)

As usual,..

Patch against 2.5.18 vanilla is available from:
ftp://ftp.kernel.org/pub/linux/kernel/people/davej/patches/2.5/

Merged patch archive: http://www.codemonkey.org.uk/patches/merged/

Check http://www.codemonkey.org.uk/Linux-2.5.html before reporting
known bugs that are also in mainline.

 -- Davej.

2.5.18-dj1
o   Fix up typos in net-sched patch from last time.	(Me)
o   More 64 bit fixes to x86-64 MTRR driver.		(Me)
o   HFS compilation fix.				(Christoph Hellwig)
o   Work around P4 Xeon errata O37			(Me)
o   Fix software delays in clps711xfb			(Me)
o   ide-scsi compile fix.				(Gert Vervoort)
o   matroxfb compile fix.				(Kai Germaschewski)
o   Remove duplication in siginfo.h			(Stephen Rothwell)
o   Remove <asm/errno.h> duplication.			(Stephen Rothwell)
o   PCMCIA config.in formatting fix.			(Adrian Bunk)
o   OPL3 compile fix.					(Takashi Iwai)
o   Fix potential oops in VFAT.				(OGAWA Hirofumi)
o   Region handling fixes in numerous drivers.		(Felipe W.Damasio,
							 William Stinson)
o   Fix broken tulip config.in				(Alex Riesen)
o   Stage 1 of LVM vg_t/lv_t sanitation.		(Anders Gustafsson)
o   vm86 page table ops locking fix.			(Benjamin LaHaise)
o   Remove unused var from v86 mode struct.		(Kasper Dupont)
o   Remove cacheline bouncing of apic_timer_irqs	(Ravikiran G Thirumalai)
o   Small cleanups in i386 head.S			(Brian Gerst)
o   Kill change_floppy warning when no floppy compiled.	(Peter Chubb)
o   semctl SUSv2 compliance.				(Christopher Yeoh)
o   Software suspend compile fix.			(Skip Ford)

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

Re: Linux 2.5.18-dj1

[A Guy Called Tyketto]

On Sun, May 26, 2002 at 02:44:39AM +0100, Dave Jones wrote:
> Resync against 2.5.18, and go through the pending patches folder.
> Mostly compilation fixes, but a handful of useful bits too.
> A lot of the pending stuff doesn't apply any more, so it may
> take a while for me to get around to applying. (ie, resending
> resynced versions of patches you sent me may be quicker)
> 
>  -- Davej.

        I seem to be bringing a shitload of bad news to the list. :) Anywho, 
got this with booting -dj1. Oops was handwritten, and thrown through ksymoops:

ksymoops 2.4.5 on i686 2.5.7.  Options used
     -v /usr/src/linux/vmlinux (specified)
     -K (specified)
     -l /proc/modules (default)
     -o /lib/modules/2.5.18/ (specified)
     -m /usr/src/linux/System.map (specified)

No modules in ksyms, skipping objects
No ksyms, skipping lsmod
Unable to handle kernel NULL pointer dereference at virtual address 00000040
c01cbd2d
*pde = 00000000
Oops: 0002
CPU:    0
EIP:    0010:[<c01cbd2d>]  Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010212
eax: 00000020   ebx: 00000000     ecx: 00000008       edx: 00000040
esi: c02f1a20   edi: 00000040     ebp: 00000020       esp: c1525d30
ds: 0010   es: 0018   ss: 0018
Stack: dfe0e928 00000000 dfe0e800 00000000 00000040 c01d086b c02794ac dfe0e928
       00000000 00000010 00000010 00000000 00000010 c0290c84 c02794ac 00000001
       00000000 dfe0e800 00007f80 c02ed6a0 c150e3c0 00000000 c1503ec0 00000000
Call Trace: [<c01d086b>] [<c01cd0b6>] [c01d36cd>] [<c01cebe8>] [<c01cec03>]
   [<c0192025>] [<c0193a94>] [<c01cba91>] [<c01c9ca7>] [<c0192435>] [<c010507e>]
   [<c0105668>]
Code: f3 a5 a8 02 74 08 66 a5 a8 01 74 01 a4 eb 0f 8d 74 26 00 55


>>EIP; c01cbd2d <fb_copy_cmap+9d/2b0>   <=====

>>esi; c02f1a20 <palette_red+0/20>
>>esp; c1525d30 <END_OF_CODE+123244c/????>

Trace; c01d086b <gen_set_cmap+8b/a0>
Trace; c01cd0b6 <fbcon_setup+866/8e0>
Trace; c0192025 <update_screen+25/60>
Trace; c0193a94 <take_over_console+e4/190>
Trace; c01cba91 <register_framebuffer+101/150>
Trace; c01c9ca7 <vgacon_cursor+1b7/1c0>
Trace; c0192435 <poke_blanked_console+55/60>
Trace; c010507e <init+1e/160>
Trace; c0105668 <kernel_thread+28/40>

Code;  c01cbd2d <fb_copy_cmap+9d/2b0>
0000000000000000 <_EIP>:
Code;  c01cbd2d <fb_copy_cmap+9d/2b0>   <=====
   0:   f3 a5                     repz movsl %ds:(%esi),%es:(%edi)   <=====
Code;  c01cbd2f <fb_copy_cmap+9f/2b0>
   2:   a8 02                     test   $0x2,%al
Code;  c01cbd31 <fb_copy_cmap+a1/2b0>
   4:   74 08                     je     e <_EIP+0xe> c01cbd3b <fb_copy_cmap+ab/2b0>
Code;  c01cbd33 <fb_copy_cmap+a3/2b0>
   6:   66 a5                     movsw  %ds:(%esi),%es:(%edi)
Code;  c01cbd35 <fb_copy_cmap+a5/2b0>
   8:   a8 01                     test   $0x1,%al
Code;  c01cbd37 <fb_copy_cmap+a7/2b0>
   a:   74 01                     je     d <_EIP+0xd> c01cbd3a <fb_copy_cmap+aa/2b0>
Code;  c01cbd39 <fb_copy_cmap+a9/2b0>
   c:   a4                        movsb  %ds:(%esi),%es:(%edi)
Code;  c01cbd3a <fb_copy_cmap+aa/2b0>
   d:   eb 0f                     jmp    1e <_EIP+0x1e> c01cbd4b <fb_copy_cmap+bb/2b0>
Code;  c01cbd3c <fb_copy_cmap+ac/2b0>
   f:   8d 74 26 00               lea    0x0(%esi,1),%esi
Code;  c01cbd40 <fb_copy_cmap+b0/2b0>
  13:   55                        push   %ebp

Kernel panic: Attempted to kill init!

        Video card used, is an ATI Rage Fury Pro (Rage128 chipset). Blew up 
right after the framebuffer kicked in. revelant part of .config:

CONFIG_AGP=y
CONFIG_AGP_VIA=y
CONFIG_DRM=y
CONFIG_DRM_R128=y
CONFIG_DRM_RADEON=y
CONFIG_FB=y
CONFIG_DUMMY_CONSOLE=y
CONFIG_VIDEO_SELECT=y
CONFIG_FB_ATY=y
CONFIG_FB_ATY_CT=y
CONFIG_FB_RADEON=y
CONFIG_FB_ATY128=y


                                                        BL.        
-- 
Brad Littlejohn                         | Email:        tyketto@wizard.com
Unix Systems Administrator,             |           tyketto@ozemail.com.au
Web + NewsMaster, BOFH.. Smeghead! :)   |   http://www.wizard.com/~tyketto
  PGP: 1024D/E319F0BF 6980 AAD6 7329 E9E6 D569  F620 C819 199A E319 F0BF

Re: Linux 2.5.18-dj1

[Russell King]

On Sun, May 26, 2002 at 12:53:14AM -0700, A Guy Called Tyketto wrote:
> >>EIP; c01cbd2d <fb_copy_cmap+9d/2b0>   <=====

This oops has existed over several 2.5 versions.  The following was thrown
around to fix it (I don't remember who though).

--- orig/drivers/video/fbcmap.c	Fri May  3 11:12:44 2002
+++ linux/drivers/video/fbcmap.c	Fri May 10 19:39:38 2002
@@ -150,9 +150,9 @@
     else
 	tooff = from->start-to->start;
     size = to->len-tooff;
-    if (size > from->len-fromoff)
+    if (size > (int)(from->len-fromoff))
 	size = from->len-fromoff;
-    if (size < 0)
+    if (size <= 0)
 	return;
     size *= sizeof(u16);
     


-- 
Russell King (rmk@arm.linux.org.uk)                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html

Re: Linux 2.5.18-dj1

[Paul Mackerras]

Russell King writes:

> This oops has existed over several 2.5 versions.  The following was thrown
> around to fix it (I don't remember who though).
> 
> --- orig/drivers/video/fbcmap.c	Fri May  3 11:12:44 2002
> +++ linux/drivers/video/fbcmap.c	Fri May 10 19:39:38 2002
> @@ -150,9 +150,9 @@
>      else
>  	tooff = from->start-to->start;
>      size = to->len-tooff;
> -    if (size > from->len-fromoff)
> +    if (size > (int)(from->len-fromoff))
>  	size = from->len-fromoff;
> -    if (size < 0)
> +    if (size <= 0)
>  	return;
>      size *= sizeof(u16);

That looks like my patch.  It stops the oops (by fixing a signed
vs. unsigned comparison bug) but doesn't fix the colormap handling.
On my G4 powerbook, which has an ATI Rage 128 (Mobility M3 AGP 2x) the
screen colours are all totally wrong under X when I boot 2.5.18.  I
have Option "UseFBDev" in my /etc/X11/XF86Config-4, so the X server is
using the kernel frame buffer driver to do colormap updates.  (The
colours are all OK on the text consoles though.)

Paul.

Re: Linux 2.5.18-dj1

[James Simmons]

> --- orig/drivers/video/fbcmap.c	Fri May  3 11:12:44 2002
> +++ linux/drivers/video/fbcmap.c	Fri May 10 19:39:38 2002
> @@ -150,9 +150,9 @@
>      else
>  	tooff = from->start-to->start;
>      size = to->len-tooff;
> -    if (size > from->len-fromoff)
> +    if (size > (int)(from->len-fromoff))
>  	size = from->len-fromoff;
> -    if (size < 0)
> +    if (size <= 0)
>  	return;
>      size *= sizeof(u16);

I going to push the fix very soon to linus. I have a bunch of new updates. 

Re: Linux 2.5.18-dj1

[James Simmons]

> That looks like my patch.  It stops the oops (by fixing a signed
> vs. unsigned comparison bug) but doesn't fix the colormap handling.
> On my G4 powerbook, which has an ATI Rage 128 (Mobility M3 AGP 2x) the
> screen colours are all totally wrong under X when I boot 2.5.18.  I
> have Option "UseFBDev" in my /etc/X11/XF86Config-4, so the X server is
> using the kernel frame buffer driver to do colormap updates.  (The
> colours are all OK on the text consoles though.)

Where is the latest ATI 128 driver? I like to port it over to the new api.