Message-id: <200903140031.09392.gene.heskett@gmail.com>
Date:	Sat, 14 Mar 2009 00:31:09 -0400
From:	Gene Heskett <gene.heskett@...il.com>
To:	Robert Hancock <hancockrwd@...il.com>
Cc:	Robin.Laing@...c-rddc.gc.ca, LKML <linux-kernel@...r.kernel.org>,
	David Newall <davidn@...idnewall.com>, yhlu.kernel@...il.com
Subject: Re: New ASUS 1701 bios for M2N SLI DELUXE

On Friday 13 March 2009, Robert Hancock wrote:
>Gene Heskett wrote:
>> Hi Robin, David and lkml list;
>>
>> I said I would report.
>>
>> I just reinstalled the 1502 version bios after spending the last 2 days
>> trying to get an hours worth of uptime without an oops.  Gave up.
>>
>> David Newell and I have been trying to find the cause of the oops, but
>> when the compile instructions David is sending me don't work, its a bit
>> difficult to troubleshoot beyond renaming the function just to see if
>> the oops follows the rename, which it does.  And with the boot girations
>> to get a working radeonhd driver now broken again, apparently by the
>> 'make mrproper' that David had me do, I'm now stuck on issue drivers for
>> drm, radeon, and radeonhd and those are noticably slower.
>>
>> So I'm back on the 1502 version of the bios, it does an oops as I sent
>> before right at entering vmlinuz, which marks me tainted, but the machine
>> is dead stable after that.
>>
>> Here is another snip of that to refresh memories:
>> [    0.000000] DMI 2.4 present.
>> [    0.000000] Phoenix BIOS detected: BIOS may corrupt low RAM, working it
>> around. [    0.000000] last_pfn = 0x120000 max_arch_pfn = 0x1000000
>> [    0.000000] x86 PAT enabled: cpu 0, old 0x7040600070406, new
>> 0x7010600070106 [    0.000000] ------------[ cut here ]------------
>> [    0.000000] WARNING: at arch/x86/kernel/cpu/mtrr/generic.c:404
>> generic_get_mtrr+0xea/0x120() [    0.000000] mtrr: your BIOS has set up an
>> incorrect mask, fixing it up. [    0.000000] Modules linked in:
>> [    0.000000] Pid: 0, comm: swapper Not tainted 2.6.28.7 #7
>> [    0.000000] Call Trace:
>> [    0.000000]  [<c042858f>] warn_slowpath+0x6f/0x90
>> [    0.000000]  [<c05193b0>] vsnprintf+0x3c0/0x7e0
>> [    0.000000]  [<c0627a00>] panic+0x15/0xee
>> [    0.000000]  [<c041a78c>] pat_init+0x7c/0xa0
>> [    0.000000]  [<c040f9fc>] post_set+0x1c/0x50
>> [    0.000000]  [<c0733f35>] dmi_string_nosave+0x4c/0x6d
>> [    0.000000]  [<c0441031>] up+0x11/0x40
>> [    0.000000]  [<c040f7ea>] generic_get_mtrr+0xea/0x120
>> [    0.000000]  [<c071f91f>] mtrr_trim_uncached_memory+0x7d/0x374
>> [    0.000000]  [<c042e583>] request_resource+0xa3/0x150
>> [    0.000000]  [<c0627af0>] printk+0x17/0x1f
>> [    0.000000]  [<c071be82>] e820_end_pfn+0xb5/0xd3
>> [    0.000000]  [<c0719fc9>] setup_arch+0x501/0xb68
>> [    0.000000]  [<c0428d89>] release_console_sem+0x189/0x1d0
>> [    0.000000]  [<c071d027>] reserve_early_overlap_ok+0x3f/0x47
>> [    0.000000]  [<c07138a4>] start_kernel+0x58/0x314
>> [    0.000000] ---[ end trace 4eaa2a86a8e2da22 ]---
>
>That's not an oops, it's a warning. Those do normally taint the kernel.
>I don't think this should really be a WARN, IMHO, as it's a BIOS bug and
>not the kernel's fault, and it's fixed up the problem. CCing Yinghai Lu,
>which it looks like wrote this warning.

I agree, the fix it does is solid.  For the later bios releases, is it 
possible that the checks are incomplete, and the later bios sneaks a broken 
map past mtrr somehow?  My nearly 60 years of troubleshooting stuff with 
circuits in it says that's the best reason I can come up with.  This, for me 
is somewhat like trying to nail jelly to a tree in an environment this 
complex.

>> So based on that, I'll now go build a 2.6.29-rc8 and see how that runs.

And its running nicely so far, about 5:55 in uptime, but I turned off all the 
selinux stuff, 10 thousand warnings about fetchmail popping up at 90 second 
intervals just got old, and Daniel doesn't seem to fix it in 2 more fixes so 
far.

>> The biggest problem with the 2.6.29 series is that apparently, for
>> security reasons, they are now doing a PHY disable in a graceful shutdown,
>> which none of the previous kernels knows how to re-enable.
>>
>> So to reboot to the 2.6.28.7 stable, you have to use the front panel reset
>> button to reboot or you will not have any onboard ethernet until you do a
>> full, pull ALL the power plugs for at least 30 seconds (I go make a cup of
>> tea, about 3 minutes) to reset the PHY's back to operational status. TBT,
>> the reset button is easier.
>>
>> Frankly, that seems like a thoroughly busted security idea, but I suppose
>> we're stuck with it.
>
>I doubt it's for security reasons. Could be due to power management or
>suspend/resume changes?

No idea, other than its a PIMA. :-)  And if for suspend/resume reasons, with a 
WOL setup, it certainly kills any chance of WOL working.

>> It also seems to me, that to mark my kernel as tainted over a fix-up that
>> makes the system dead stable, is executing the messenger.  IMNSHO, its
>> ASUS who ought to be shot for not having any method of filing a bug report
>> against their crappy bios.  3 emails sent to the only address I was able
>> to find for ASUS have had the same effect as sending them to /dev/null.
>>
>> If anyone on the LKML knows how to contact ASUS, please advise them that
>> there is at least one VERY unhappy camper/user of the
>> $285 USD M2N SLI DELUXE motherboard, I'm sorry I ever laid eyes on it.
>> The newly released version 1701 bios (it unzips as
>> M2N-SLI-Deluxe-1701.BIN) still doesn't get it right AFAIAC.  The kernel
>> writers at least know what to do about the older version bios.  Maybe ASUS
>> coded the new bios to get past your test, but is it still fscked?  That is
>> certainly my opinion...
>>
>> Thanks everybody.  Now back to your regularly scheduled programming. :)

Repeat, thanks everybody.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
He hadn't a single redeeming vice.
		-- Oscar Wilde

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists