linux-mips
[Top] [All Lists]

Re: [PATCH] Further TLB handler optimizations

To: Rojhalat Ibrahim <ibrahim@schenk.isar.de>
Subject: Re: [PATCH] Further TLB handler optimizations
From: Thiemo Seufer <ica2_ts@csv.ica.uni-stuttgart.de>
Date: Tue, 11 Jan 2005 19:04:40 +0100
Cc: linux-mips@linux-mips.org
In-reply-to: <41E40E8D.60600@schenk.isar.de>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <20041224085645.GJ3539@rembrandt.csv.ica.uni-stuttgart.de> <20050107190605.GG31335@rembrandt.csv.ica.uni-stuttgart.de> <41E27A6A.5060204@schenk.isar.de> <20050110140429.GC15344@rembrandt.csv.ica.uni-stuttgart.de> <41E29DF5.6040800@schenk.isar.de> <20050110154246.GH15344@rembrandt.csv.ica.uni-stuttgart.de> <41E38A0A.1010507@schenk.isar.de> <41E394A0.30205@schenk.isar.de> <20050111162646.GA31149@rembrandt.csv.ica.uni-stuttgart.de> <41E40E8D.60600@schenk.isar.de>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.5.6+20040907i
Rojhalat Ibrahim wrote:
[snip]
> >>Well, at least for a 32-bit kernel. If I compile a 64-bit
> >>kernel it still stops when it should start init. Any ideas?
> >
> >Is this really related to my TLB exception handler patch? The 64bit
> >versions of it are pretty well tested now, so it would be most likely
> >some flaw specific to your CPU. The 64bit kernel had no optimized TLB
> >handlers before.
> 
> Don't know. It's just that it worked before. But I might just
> have been lucky then. I just checked again and found that without
> SMP enabled 64 bit is working too. So the problem might indeed
> not be related to your patch.

The pte updates for SMP are tricky but the same for 32 and 64 bit.
The old 32bit SMP code simply ignored the race condition, the 64bit SMP
version had no optimization but called always do_fage_fault.

64bit SMP systems known to work so far are SGI IP27 and Broadcom SB1250.


Thiemo

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