Linux 2.5.18-dj1

DISCUSSION Dave Jones 5 participants 6 messages May 26, 2002 – May 28, 2002

Share & tools

Dave Jones

author

2002-05-26 01:43 UTC #0 lore raw

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)

Signature · 2 lines

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

A Guy Called Tyketto

2002-05-26 07:53 UTC #1 lore raw

Quoted 7 lines · 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)
-- 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

Quoted 4 lines

Quoted 4 lines
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.

Signature · 4 lines

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

Russell King

2002-05-26 08:00 UTC #2 lore raw

In reply to Dave Jones

Quoted 1 line · A Guy Called Tyketto

Quoted 1 line
Quoted 1 line
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).

drivers/video/fbcmap.c +2 -2

@@ -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);

Signature · 2 lines

Russell King ([email protected])                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html

Paul Mackerras

2002-05-26 11:43 UTC #3 lore raw

Quoted 6 lines · Russell King

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

drivers/video/fbcmap.c line 157

Quoted 10 lines

drivers/video/fbcmap.c +2 -2

@@ -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.

James Simmons

2002-05-28 17:33 UTC #4 lore raw

Quoted 8 lines

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.

James Simmons

2002-05-28 17:31 UTC #5 lore raw

In reply to Russell King

Quoted 4 lines

drivers/video/fbcmap.c line 157

Quoted 10 lines

drivers/video/fbcmap.c +2 -2

@@ -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.