[Top] [All Lists]

Re: SMTC support status in latest git head.

To: Anoop P A <>
Subject: Re: SMTC support status in latest git head.
From: "Kevin D. Kissell" <>
Date: Tue, 28 Dec 2010 00:43:10 -0800
Cc: STUART VENTERS <>, "Anoop P.A." <>,
Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=default;; h=Received:Message-ID:Date:From:User-Agent:MIME-Version:To:CC:Subject:References:In-Reply-To:Content-Type:Content-Transfer-Encoding:X-Source:X-Source-Args:X-Source-Dir; b=SGvJfbZet55+DFum6TJwur6YsuvMZAB9CEuazk5s4uYKOjRznMriDVIF660aAq4R20YOS1HP6vE+D7xSSbD6ZtoVTOZ7pIP+5ugM6Cv8KeXrRRQzLnrXFF6aMZliAhRb;
In-reply-to: <1293524389.27661.210.camel@paanoop1-desktop>
Original-recipient: rfc822;
References: <> <1293470392.27661.202.camel@paanoop1-desktop> <1293524389.27661.210.camel@paanoop1-desktop>
User-agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv: Gecko/20101207 Thunderbird/3.1.7
I took a quick look last night, and the only thing that looked vaguely dangerous in changes since the timer changes I alluded to earlier was the global naming cleanup of irq-related function names that David Howell submitted. The diff didn't look dangerous in itself, but some of the definitions are nested subtly for SMTC to maximize the amount of common code, and I could imagine something getting lost in translation there. If that were really the problem, it would of course affect much more than just the timer subsystem, but early in the boot process, timers are pretty much the only interrupts that have to be handled correctly.

I'm travelling today, but will take a look at timekeeping_notify() tomorrow or the next day...


On 12/28/10 12:19 AM, Anoop P A wrote:

I had a glance into the code diff without notice of any suspect-able
code .
Tracing the hang showed that it is getting hanged in timekeeping_notify


PS: I may not be available until Thursday

On Mon, 2010-12-27 at 22:49 +0530, Anoop P A wrote:
Hi Kevin,

It is very unlikely that the patch you pointed has any impact on the the
hang I am seeing. The patch you have mentioned got into kernel around
2.6.32 timeframe. I am able to boot both 2.6.32 and  2.6.33 kernel ( +
stackframe patch) .

Hi Stuart,

I haven't got much time to spend on this today.

I had got 2.6.36-stable(+ stack frame patch) booting last day and I have
observed hang issue with 2.6.37-rc1 ( Same as rc6 and current git head)

So probably some patches in 2.6.37 branch introduced this hang.

Hopefully I will get some free slot tomorrow so that I can look into
code diff .


On Mon, 2010-12-27 at 09:49 -0600, STUART VENTERS wrote:

Outstanding, sometimes it's better to be lucky than good.


Maybe we can get lucky again.

If you can isolate the .33 works/.37 works_not bug to a specific pair of 
    I'll be happy to do another diff.

Hope you'll have had a good Christmas as well.
   We've had snow in Alabama since Christmas eve!



-----Original Message-----
From: Kevin D. Kissell []
Sent: Friday, December 24, 2010 5:34 PM
To: Anoop P A
Subject: Re: SMTC support status in latest git head.

Ah, well, at least we have a stackframe.h fix that preserves David's
performance tweak for the deeper pipelined processors.  In looking for
this, I did notice that someone did some modification to the SMTC clock
tick logic that I was skeptical had ever been tested.  If you've still
got that kernel binary handy, you might check to see if it boots with
maxtcs=1 maxvpes=1, maxtcs=2 maxvpes=1, and/or maxtcs=2 maxvpes=2.

Oh, yes, and Merry Christmas one and all!


              Kevin K.

On 12/24/10 8:02 AM, Anoop P A wrote:
On Fri, 2010-12-24 at 06:53 -0800, Kevin D. Kissell wrote:
Excellent!  Now, does the attached patch (relative to also
fix things, while preserving the other fixes and performance enhancements?

I have tested that patch with 2.6.37 branch it well passes calibration
loop but hangs after switching to mips closource

TC 6 going on-line as CPU 6
Brought up 7 CPUs
bio: create slab<bio-0>   at 0
SCSI subsystem initialized
Switching to clocksource MIPS

I Presume this is a different issue as restoring older file didn't help
much to get rid of this hang.

diff --git a/arch/mips/include/asm/stackframe.h
index 58730c5..7fc9f10 100644
--- a/arch/mips/include/asm/stackframe.h
+++ b/arch/mips/include/asm/stackframe.h
@@ -195,9 +195,9 @@
                 * to cover the pipeline delay.
                .set    mips32
-               mfc0    v1, CP0_TCSTATUS
+               mfc0    v0, CP0_TCSTATUS
                .set    mips0
-               LONG_S  v1, PT_TCSTATUS(sp)
+               LONG_S  v0, PT_TCSTATUS(sp)
   #endif /* CONFIG_MIPS_MT_SMTC */
                LONG_S  $4, PT_R4(sp)
                LONG_S  $5, PT_R5(sp)


On 12/24/10 6:39 AM, Anoop P A wrote:
Hi Kevin, Stuart ,

Woohooo You guys spotted !.;a=commit;h=d5ec6e3c seems to be
the culprit

Once I restored previous version of stackframe.h 2.6.33-stable started
booting !.


On Fri, 2010-12-24 at 04:32 -0800, Kevin D. Kissell wrote:
Thank you, Stuart!  I've spotted some definite breakage to SMTC between
those versions.  In arch/mips/include/asm/stackframe.h, someone moved
the store of the Status register value in SAVE_SOME (line 169 or 204,
depending on the version) from two instructions after the mfc0 to a
point after the #ifdef for SMTC, presumably to get better pipelining of
the register access.  Unfortunately, the v1 register is also used in the
SMTC-specific fragment to save TCStatus, so the Status value gets
clobbered before it gets stored.  This will eventually result in the
Status register getting a TCStatus value, which has some bits on common,
but isn't identical and sooner or later Bad Things will happen.

I'm a little surprised this wasn't caught by visual inspection of the patch.

Possible solutions would include reverting the store of the CP0_STATUS
value to the block above the #ifdef, or, to retain whatever performance
advantage was obtained by moving the store downward, to use v0/$2
instead of v1/$3, as the staging register for the TCStatus value.  I'd
lean toward the second option, but I'm not in a position to test and
submit a patch just now.


                Kevin K.

On 12/23/10 1:09 PM, STUART VENTERS wrote:

I'm not sure if it's useful,
       but finally I got the time to look at the two kernel versions Anoop 
pointed out.
        works   2.6.32-stable with patch 804
        works_not 2.6.33-stable

greping for files with CONFIG_MIPS_MT_SMTC
       and looking for timer interrupt related stuff found the following 


      SAVE_SOME SAVE_TEMP get/set_saved_sp




Enclosed are the two subsets of files for a more expert look.

I'll try to look in more detail after Christmas.



<Prev in Thread] Current Thread [Next in Thread>