Linux 2.5.67-ac2

Linux 2.5.67-ac2

[Alan Cox]

Next set of updates. Cautionary words still appy.

Handle with care, no naked flames, do not inhale....

Linux 2.5.67-ac2
o	Remove now unrequired identity hack for proc	(me)
o	Resync with Linus -bk8 snapshot
o	Merge core H8300 updates			(Yoshinori Sato)
o	Merge summit ID strings				(James Cleverdon)
o	Input layer typo fixes				(Randy Dunlap)
o	Update Intermezzo maintainer info		(Jörn Engle)
o	Further C99 initializer updates for audio	(Art Haas)
o	Fix sound driver crashes with pnp		(Daniel Ritz)
o	Rio requires uaccess.h	(Bug #566)		(Bob Miller)
o	Sx requires uaccess.h				(John Kim)
o	Minor ipmi fixes				(Corey Minyard)
o	z2ram compile fix				(Geert Uytterhoeven)
o	Fix an SMP boot bug in -ac			(Osamu Tomita)
o	Unbreak visws					(Andrey Panin)
o	Switch ide-cs to pcmcia_driver			(Christoph Hellwig)
o	Fix the trackpoint problem on Dell8100		(Arne Koewing)
o	Add another broken bios to the list

Linux 2.5.67-ac1
o	Fix the problem with Serverworks bootup when	(Robert Hentosh,
	the BIOS forced PIO modes			 me)
o	BIO walking functions 			(Bartlomiej Zolnierkiewicz &
o	BIO walking code documentation		 Suparna Bhattacharya)
o	Fix ide taskfile for new prehandlers	(Bartlomiej Zolnierkiewicz)
o	Fix ide taskfile DMA read/write		(Bartlomiej Zolnierkiewicz)
o	Remove old style non taskfile PIO 	(Bartlomiej Zolnierkiewicz)
o	Clean up duplicated IDE code		(Bartlomiej Zolnierkiewicz)
o	Huge update to DVB 				(Michael Hunold)
	| This reconciles a large fork in DVB development
o	Fix build for PC164 Alpha			(Marc Zyngier)
o	Clean up PCI_INTERRUPT_LINE in IDE		(me)
o	Fix mdelay problem on PPC			(Paul Mackerras)
o	Add pc9800.h header				(Osamu Tomita)
o	Add pc9800 uni map files (not console code)	(Osamu Tomita)
o	Resync 2.4 and 2.5 opl3sa2 support		(Daniel Ritz)
o	Fix missing include for copy*user in cpufreq	(me)
o	Fix slip for new module locking on ldisc	(me)
o	Update lp486e to new style locking etc		(me)
o	Update 3c574_cs to new style locking		(me)
o	Clean up scc locking for 2.5			(me)
o	Merge 2.5.67
o	Fix mtd build					(me)
o	FIx 3c505 build					(me)

Linux 2.5.66-ac2
o	Resync with Linus -bk8
o	Fix modules build				(me)
o	Fix arch syscalls to return long		(Randy Dunlap)
o	USBfs kerneldoc					(David Brownell)
o	More i2c updates				(Greg Kroah Hartmann)
	| FIxes several driver compiles
o	Fix IDE locking/phase handling on timeout	(Manfred Spraul)
o	C99 initialisers for DMAsound			(Maciej Soltysiak)
o	C99 initialisers for OSS audio			(Maciej Soltysiak)
o	C99 intialisers for emu10k1			(Maciej Soltysiak)
o	Move dead MOD_ calls from floppy driver		(Bob Miller)
o	Fix macmace abuse of GFP_DMA			(Matthew Wilcox)
o	Improve hpt kconfig entry			(Adrian Bunk)
o	Generic HDLC updates				(Krzysztof Halasa)
o	Fix es968 kmalloc parameters			(Pablo Meinchini)
o	Fix als100 kmalloc parameters			(Pablo Meinchini)
o	PC98xx updates to existing merge		(Osamu Tomita)
	| Keyboard, ALSA, floppy
o	Fix misc_register fail path for upd4990a	(Stephan Maciej)
o	Fix pegasus endian bug				(Paul Mackerras)
o	Wireless needs __init				(Paul Mackerras)
o	Fix gus compilation when built in (lock		(Peter Waechtler)
	clash)
o	Sony PI driver update				(Stelian Pop)
o	Spelling fixes for Alpha			(Steven Cole)
o	Spelling fixes for x86-64			(Steven Cole)
o	Add a "blank now" key mapping			(Pavel Machek)
o	Remove 23 bogus includes of version.h		(Burton Windle)
o	Fix cs4232 build				(Daniel Ritz)
o	Update v850 architecture			(Miles Bader)
o	Fix taint mishandling for AMD CPU		(Manfred Spraul)
o	Fix compile of dt019x audio			(John Kim)
o	Kill "compatmac"				(Adrian Bunk)
o	Fix via82cxxx_audio build			(me)
o	Fix acpi build					(me)
o	Resync with bk9
o	Fix oprofile build				(John Levon)
o	PnP updates					(Adam Belay)
o	Fix visws framebuffer compile			(Andrey Panin)
o	Avoid LBA48 modes on disks that don't need	(Jens Axboe)
	them (saves a 2nd command cycle on each I/O
	as suggested by Mark Lord)
o	Clean up ide list handling for drives		(Alexander Atanasov)
o	Remove present check from drivers (now		(Alexander Atanasov)
	handled by the list stuff)
o	Fix SMP boot timer oops				(me)
	| Thanks to Steven Cole for pinning this down to
	| 5 lines of change
o	Port ltpc driver to spinlocks			(me)
o	Fix cadet driver missing symbol			(me)
o	Fix cops driver locking				(me)
o	Fix arcnet.c locking				(me)

Linux 2.5.66-ac1
o	Fix up ESI handling in esp.c			(me)
	| Lots more needs fixing in this driver yet
o	Merge Linus 2.5.66
o	Fix cramfs compile problems			(Jeremy Brown, me)
o	Fix mad16 breakage				(Adrian Bunk)
o	Iphase fixes port to 2.5			(Chas Williams,
							 Eric Leblond)
o	Fix ipc/msg race				(Manfred Spraul)
o	Let hdparm know about speed change fails	(Jens Axboe)
o	Don't issue WIN_SET_MAX on older drivers	(Jens Axboe)
	(Breaks some Samsung)
o	Resync with Linus bk3 snapshot

Re: Linux 2.5.67-ac2: ide reset issue

[rain.wang]

[-- Attachment #1: Type: text/plain, Size: 1264 bytes --]

Hi,
    I tested 2.5.67-ac2 with two continuous drive reset call,
dmesged out oops messages as below. system didn't crash
like under 2.4.21-pre1-ac1, but that race still exist.
    oops message:

hda: DMA disabled
------------[ cut here ]------------
kernel BUG at drivers/ide/ide.c:1603!
invalid operand: 0000 [#1]
CPU:    0
EIP:    0060:[<c0216ac1>]    Not tainted
EFLAGS: 00010082
EIP is at generic_ide_ioctl+0x471/0x860
eax: c7e8ee60   ebx: 00000202   ecx: c038be8c   edx: c0211d10
esi: c02ba3cb   edi: c038be8c   ebp: c781df58   esp: c781df0c
ds: 007b   es: 007b   ss: 0068
Process hdparm (pid: 203, threadinfo=c781c000 task=c7bd06a0)
Stack: c7dc21c0 c122f3e0 08049cf4 00000000 00000000 08049cf4 c122f3e0 c7dc6000
       3e9f6c00 00030002
Call Trace:
 [<c0201af2>] blkdev_ioctl+0x82/0x3cc
 [<c0151d41>] sys_ioctl+0x81/0x220
 [<c0142d13>] sys_write+0x33/0x40
 [<c01090e3>] syscall_call+0x7/0xb

Code: 0f 0b 43 06 0a a3 2b c0 c7 40 08 01 00 00 00 53 9d 57 e8 a8
 ide0: reset: success

    I don't know if there's enough reason to change reset semantics
now to wait for completion, so that the next call be free of race.
and  I once had a simpler fix to let it delay another 50ms, that works
on my box but seems not a thorough one. does it help?

Regards
rain.w

[-- Attachment #2: ide.c.diff.3 --]
[-- Type: text/plain, Size: 354 bytes --]

--- /usr/src/linux-2.5.67-ac2/drivers/ide/ide.c	Fri Apr 18 10:11:01 2003
+++ ide.c	Fri Apr 18 11:14:26 2003
@@ -1608,6 +1608,10 @@
 			HWGROUP(drive)->busy = 1;
 			spin_unlock_irqrestore(&ide_lock, flags);
 			(void) ide_do_reset(drive);
+
+			/* wait for another 50ms */
+			mdelay(50);
+
 			if (drive->suspend_reset) {
 /*
  *				APM WAKE UP todo !!

Re: Linux 2.5.67-ac2

[Rudmer van Dijk]

On Friday 18 April 2003 01:19, Alan Cox wrote:
> Next set of updates. Cautionary words still appy.
> 
> Handle with care, no naked flames, do not inhale....

It compiles, but it does not boot:
kernel panic: cannot find init

this is with devfs enabled and mounted at boot.
I can't find anything in the lists although I thought I saw someone reporting 
a similar problem.
will try it without devfs

	Rudmer

Re: Linux 2.5.67-ac2: ide reset issue

[Alan Cox]

>     I don't know if there's enough reason to change reset semantics
> now to wait for completion, so that the next call be free of race.
> and  I once had a simpler fix to let it delay another 50ms, that works
> on my box but seems not a thorough one. does it help?

BWGROUP(drive)->busy should never reach zero until the reset is
done. The 50mS miught be enough that this occurs, as might waiting
for HWGROUP(drive)->busy hitting 0. I don't yet understand why it
matters, and to fix it properly I have to figure that out.

If you need reliable reset for something like a test harness, or
IDE drive tester its a usable workaround, but I need to fix it
properly (eventually)

> +			/* wait for another 50ms */
> +			mdelay(50);

In your test set is HWGROUP(drive)->busy always zero after the
mdelay ?

Re: Linux 2.5.67-ac2

[Bill Davidsen]

On Fri, 18 Apr 2003, Rudmer van Dijk wrote:

> On Friday 18 April 2003 01:19, Alan Cox wrote:
> > Next set of updates. Cautionary words still appy.
> > 
> > Handle with care, no naked flames, do not inhale....
> 
> It compiles, but it does not boot:
> kernel panic: cannot find init
> 
> this is with devfs enabled and mounted at boot.
> I can't find anything in the lists although I thought I saw someone reporting 
> a similar problem.
> will try it without devfs

Let us know, my memory is that Alan doesn't use devfs and it's "lightly
tested" if I remember his term.

-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

Re: Linux 2.5.67-ac2: ide reset issue

[rain.wang]

Alan Cox wrote:

> >     I don't know if there's enough reason to change reset semantics
> > now to wait for completion, so that the next call be free of race.
> > and  I once had a simpler fix to let it delay another 50ms, that works
> > on my box but seems not a thorough one. does it help?
>
> BWGROUP(drive)->busy should never reach zero until the reset is
> done. The 50mS miught be enough that this occurs, as might waiting
> for HWGROUP(drive)->busy hitting 0. I don't yet understand why it
> matters, and to fix it properly I have to figure that out.
>
> If you need reliable reset for something like a test harness, or
> IDE drive tester its a usable workaround, but I need to fix it
> properly (eventually)
>

I agree. I found the reason seems some strange there. reset call set
a 50ms's wait handler and return to user at once, when succeed in
the first poll and handler return, there's always about another 50ms
needed to cleanup the path(I once tested values lager and smaller
than 50ms and found about 48ms needed at least on my box).  so
the following call would race it if there's no such a delay, although
there's actually few chances to do continuous reset call, I thought.

> > +                     /* wait for another 50ms */
> > +                     mdelay(50);
>
> In your test set is HWGROUP(drive)->busy always zero after the
> mdelay ?

I think it is.