2.5.69: VIA IDE still broken

2.5.69: VIA IDE still broken

[Felix von Leitner]

I can't believe this still isn't fixed!

 hda: dma_timer_expiry: dma status == 0x24
 hda: lost interrupt
 hda: dma_intr: bad DMA status (dma_stat=30)
 hda: dma_intr: status=0x50 { DriveReady SeekComplete }

 hda: dma_timer_expiry: dma status == 0x24
 hda: lost interrupt
 hda: dma_intr: bad DMA status (dma_stat=30)
 hda: dma_intr: status=0x50 { DriveReady SeekComplete }

My hda is in perfect health and this does not happen on the same
hardware with 2.4.* or 2.5.63.  I reported this before and got the
answer that to fix this, recent changes in the IDE code would have to be
reverted.  Apparently I was unreasonably hasty in assuming that that
would be done now that the need to do it has been established.

I would appreciate it if the fix would be integrated into 2.5.70.

Amazing: the only hardware components in my machine that actually work
as expected with recent Linux 2.5 kernels are the network cards, the RAM
and the keyboard, and I had to replace a tulip card with an eepro100 for
that.  Even the CPU appears to run too hot with Linux, causing the
system to boot spontaneously under load, and because ACPI is terminally
broken in Linux and has been every time I tried it, I can't do much
about it.  Firewire does not like me (modprobe eth1394 -> oops), IDE
loses interrupts (see above), my USB mouse stops working as soon as I
plug in my USB hard disk (which works fine on my notebook and under
Windows), using my IDE CD-R causes the machine to freeze while cdrecord
does OTP, finalizing or eject.  The nvidia graphics card takes major
patching to work at all with X, and all of these components are
well-known brand components from tier 1 suppliers that were chosen for
reliability and market penetration over price.  I envy people who can
still evangelize Linux under circumstances like this.  I sure as hell
can not.

Felix

Re: 2.5.69: VIA IDE still broken

[Alan Cox]

On Iau, 2003-05-08 at 23:09, Felix von Leitner wrote:
> hardware with 2.4.* or 2.5.63.  I reported this before and got the
> answer that to fix this, recent changes in the IDE code would have to be
> reverted.  Apparently I was unreasonably hasty in assuming that that
> would be done now that the need to do it has been established.

Except that the change that triggers this seems correct, so until
someone explains the problem it won't.

> that.  Even the CPU appears to run too hot with Linux, causing the
> system to boot spontaneously under load, 

Sounds like your hardware is faulty. 

Re: 2.5.69: VIA IDE still broken

[Bartlomiej Zolnierkiewicz]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 2052 bytes --]


Apply patch.

On Fri, 9 May 2003, Felix von Leitner wrote:

> I can't believe this still isn't fixed!
>
>  hda: dma_timer_expiry: dma status == 0x24
>  hda: lost interrupt
>  hda: dma_intr: bad DMA status (dma_stat=30)
>  hda: dma_intr: status=0x50 { DriveReady SeekComplete }
>
>  hda: dma_timer_expiry: dma status == 0x24
>  hda: lost interrupt
>  hda: dma_intr: bad DMA status (dma_stat=30)
>  hda: dma_intr: status=0x50 { DriveReady SeekComplete }
>
> My hda is in perfect health and this does not happen on the same
> hardware with 2.4.* or 2.5.63.  I reported this before and got the
> answer that to fix this, recent changes in the IDE code would have to be
> reverted.  Apparently I was unreasonably hasty in assuming that that
> would be done now that the need to do it has been established.
>
> I would appreciate it if the fix would be integrated into 2.5.70.
>
> Amazing: the only hardware components in my machine that actually work
> as expected with recent Linux 2.5 kernels are the network cards, the RAM
> and the keyboard, and I had to replace a tulip card with an eepro100 for
> that.  Even the CPU appears to run too hot with Linux, causing the
> system to boot spontaneously under load, and because ACPI is terminally
> broken in Linux and has been every time I tried it, I can't do much
> about it.  Firewire does not like me (modprobe eth1394 -> oops), IDE
> loses interrupts (see above), my USB mouse stops working as soon as I
> plug in my USB hard disk (which works fine on my notebook and under
> Windows), using my IDE CD-R causes the machine to freeze while cdrecord
> does OTP, finalizing or eject.  The nvidia graphics card takes major
> patching to work at all with X, and all of these components are
> well-known brand components from tier 1 suppliers that were chosen for
> reliability and market penetration over price.  I envy people who can
> still evangelize Linux under circumstances like this.  I sure as hell
> can not.
>
> Felix

So what are you waiting for? Stop whining and start hacking now!
--
Bartlomiej

[-- Attachment #2: Type: TEXT/PLAIN, Size: 1124 bytes --]


# Fix masked_irq arg handling for ide_do_request().
# Solves "hdx: lost interrupt" bug.
#
# Bartlomiej Zolnierkiewicz <bzolnier@elka.pw.edu.pl>

 drivers/ide/ide-io.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff -puN drivers/ide/ide-io.c~ide_masked_irq drivers/ide/ide-io.c
--- linux-2.5.69/drivers/ide/ide-io.c~ide_masked_irq	Thu May  8 17:16:27 2003
+++ linux-2.5.69-root/drivers/ide/ide-io.c	Thu May  8 17:16:27 2003
@@ -850,14 +850,14 @@ queue_next:
 		 * happens anyway when any interrupt comes in, IDE or otherwise
 		 *  -- the kernel masks the IRQ while it is being handled.
 		 */
-		if (hwif->irq != masked_irq)
+		if (masked_irq != IDE_NO_IRQ && hwif->irq != masked_irq)
 			disable_irq_nosync(hwif->irq);
 		spin_unlock(&ide_lock);
 		local_irq_enable();
 			/* allow other IRQs while we start this request */
 		startstop = start_request(drive, rq);
 		spin_lock_irq(&ide_lock);
-		if (hwif->irq != masked_irq)
+		if (masked_irq != IDE_NO_IRQ && hwif->irq != masked_irq)
 			enable_irq(hwif->irq);
 		if (startstop == ide_released)
 			goto queue_next;

_

Re: 2.5.69: VIA IDE still broken

[Greg KH]

On Fri, May 09, 2003 at 12:09:10AM +0200, Felix von Leitner wrote:
> Amazing: the only hardware components in my machine that actually work
> as expected with recent Linux 2.5 kernels are the network cards, the RAM
> and the keyboard, and I had to replace a tulip card with an eepro100 for
> that.  Even the CPU appears to run too hot with Linux, causing the
> system to boot spontaneously under load, and because ACPI is terminally
> broken in Linux and has been every time I tried it, I can't do much
> about it.  Firewire does not like me (modprobe eth1394 -> oops), IDE
> loses interrupts (see above), my USB mouse stops working as soon as I
> plug in my USB hard disk (which works fine on my notebook and under
> Windows), using my IDE CD-R causes the machine to freeze while cdrecord
> does OTP, finalizing or eject.  The nvidia graphics card takes major
> patching to work at all with X, and all of these components are
> well-known brand components from tier 1 suppliers that were chosen for
> reliability and market penetration over price.  I envy people who can
> still evangelize Linux under circumstances like this.  I sure as hell
> can not.

Have you filed bugs in bugzilla.kernel.org for these issues so that the
developers can know about them to fix them?

thanks,

greg k-h

Re: 2.5.69: VIA IDE still broken

[Ed Sweetman]

>Amazing: the only hardware components in my machine that actually work
>as expected with recent Linux 2.5 kernels are the network cards, the RAM
>and the keyboard, and I had to replace a tulip card with an eepro100 for
>that.  Even the CPU appears to run too hot with Linux, causing the
>system to boot spontaneously under load, and because ACPI is terminally
>broken in Linux and has been every time I tried it, I can't do much
>about it.  Firewire does not like me (modprobe eth1394 -> oops), IDE
>loses interrupts (see above), my USB mouse stops working as soon as I
>plug in my USB hard disk (which works fine on my notebook and under
>Windows), using my IDE CD-R causes the machine to freeze while cdrecord
>does OTP, finalizing or eject.  The nvidia graphics card takes major
>patching to work at all with X, and all of these components are
>well-known brand components from tier 1 suppliers that were chosen for
>reliability and market penetration over price.  I envy people who can
>still evangelize Linux under circumstances like this.  I sure as hell
>can not.

that's funny. I'm using a via board with a tulip card with acpi with usb 
and with the via ide controller and with an ide cdr and all of it works 
PERFECTLY (2.5.69 is the first 2.5 kernel which has been as good with 
all my hardware as 2.4.2x) and there are no errors generated and 
everything is working very responsively which is nice for a change.   I 
dont have anything that does firewire so i cant say much about that. I 
have an nvidia card but it's on the P4 box which uses an intel 
motherboard so it's not very useful here.  But my matrox card is running 
  better than with any other kernel according to x11perf. I also have a 
wintv card working perfectly and sblive dido.

The fact that you're having cooling problems causing reboots leads me to 
believe that this is a hardware related problem due to either poor 
construction or insufficient cooling and it's going to be really hard to 
debug problems when you have overheating problems seeing as how 
overheating causes random crap to occur. And even if you did fix the 
cooling you may have already caused irreversible damage to components on 
the motherboard or in the cpu-itself. It's easy to still back linux with 
cases like this because they're not indicative of the kernel overall or 
even in an obvious way.

I for one am extremely pleased with 2.5.69 as it is much more stable 
than 2.3.x kernels were as they approached 2.4.x.  It's user-land 
software that i'm currently fighting with as it takes userland a long 
time to catch up to the latest kernel and with big api changes like no 
more proc i2c interface i've gotta go without some programs (procmeter) 
for sensor data (which works 100% with sysfs).  That and the fact that 
userland really hates bug reports when you're using an odd numbered 
kernel (samba).

Re: 2.5.69: VIA IDE still broken

[Henning P. Schmiedehausen]

Felix von Leitner <felix-kernel@fefe.de> writes:

>reliability and market penetration over price.  I envy people who can
>still evangelize Linux under circumstances like this.  I sure as hell
>can not.

Solution A: Don't use _development_ kernels.

Solution B: Use another OS where development snapshots have production quality.

	Regards
		Henning

-- 
Dipl.-Inf. (Univ.) Henning P. Schmiedehausen          INTERMETA GmbH
hps@intermeta.de        +49 9131 50 654 0   http://www.intermeta.de/

Java, perl, Solaris, Linux, xSP Consulting, Web Services 
freelance consultant -- Jakarta Turbine Development  -- hero for hire

Re: 2.5.69: VIA IDE still broken

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 672 bytes --]

On Fri, 09 May 2003 00:09:10 +0200, Felix von Leitner <felix-kernel@fefe.de>  said:

> does OTP, finalizing or eject.  The nvidia graphics card takes major
> patching to work at all with X, and all of these components are

Umm... see http://www.minion.de/nvidia - or have you tried that and it
still doesn't work, or you trying to get the XFree driver working, or is
it some other issue?

(One gotcha that hosed *ME* up was the fact that recent XFree86 are TLS-sensitive,
and NVidia's latest few series of drivers included TLS libs for OpenGL - but
I didn't have /lib/tls and/or /usr/lib/tls in /etc/ld.so.conf so ldconfig
didn't DTRT - THAT horqued things up but good).


[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: 2.5.69: VIA IDE still broken

[Hirling Endre]

On Fri, 2003-05-09 at 00:09, Felix von Leitner wrote:

> that.  Even the CPU appears to run too hot with Linux, causing the
> system to boot spontaneously under load, and because ACPI is terminally
> broken in Linux and has been every time I tried it, I can't do much
> about it.  Firewire does not like me (modprobe eth1394 -> oops), IDE
> loses interrupts (see above), my USB mouse stops working as soon as I
> plug in my USB hard disk (which works fine on my notebook and under
> Windows), using my IDE CD-R causes the machine to freeze while cdrecord
> does OTP, finalizing or eject.  The nvidia graphics card takes major

These sound like you have a weak power supply, try putting in a bigger
one and see if it cures the problems.

-m-

Re: 2.5.69: VIA IDE still broken

[Bill Davidsen]

On Fri, 9 May 2003, Felix von Leitner wrote:

> I can't believe this still isn't fixed!
> 
>  hda: dma_timer_expiry: dma status == 0x24
>  hda: lost interrupt
>  hda: dma_intr: bad DMA status (dma_stat=30)
>  hda: dma_intr: status=0x50 { DriveReady SeekComplete }
> 
>  hda: dma_timer_expiry: dma status == 0x24
>  hda: lost interrupt
>  hda: dma_intr: bad DMA status (dma_stat=30)
>  hda: dma_intr: status=0x50 { DriveReady SeekComplete }
> 
> My hda is in perfect health and this does not happen on the same
> hardware with 2.4.* or 2.5.63.

That seems to reduce the chances of hardware problems. At the least later
kernels tickle the problem if there is one.

> hardware with 2.4.* or 2.5.63.  I reported this before and got the
> answer that to fix this, recent changes in the IDE code would have to be
> reverted.  Apparently I was unreasonably hasty in assuming that that
> would be done now that the need to do it has been established.
> 
> I would appreciate it if the fix would be integrated into 2.5.70.

I wouldn't assume that the fix is known and as simple as you seem to
think. Other people aren't having the problem. Doesn't mean there isn't
one, but the changes you want reverted were made for a reason.

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