Date: Tue, 11 Dec 2001 15:41:38 +1100 From: Peter Jeremy <peter.jeremy@alcatel.com.au> To: Jameel Akari <jakari@bithose.com> Cc: freebsd-alpha@FreeBSD.ORG Subject: Re: How to enable logging of machine checks? Message-ID: <20011211154137.A73243@gsmx07.alcatel.com.au> In-Reply-To: <Pine.OSF.4.33.0112091637370.53952-100000@poptart.bithose.com>; from jakari@bithose.com on Sun, Dec 09, 2001 at 04:44:38PM -0500 References: <20011210072424.B63990@gsmx07.alcatel.com.au> <Pine.OSF.4.33.0112091637370.53952-100000@poptart.bithose.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2001-Dec-09 16:44:38 -0500, Jameel Akari <jakari@bithose.com> wrote: >On Mon, 10 Dec 2001, Peter Jeremy wrote: >> And, based on my decoding of the LCA memory interface, the L2 cache ECC >> is disabled by default. > > Hmm... odd. Whatever messages I saw were about ECC traps, and >obviously if the cache ECC was off, the machine would have died. The Multia supports ECC on both the RAM and cache - both of which can be separately enabled/disabled. By my reading of the LCA datasheet, combined with what SRM displays as the register values, the RAM ECC is on but the cache ECC is off. Possible explanations: - Your ECC errors were RAM not cache. Though I can't explain why the 74F623 would affect this. - We are running different SRM's - I'm running BL5 V3.8-3 - I've misread the datasheet. - The LCA doesn't match the datasheet. For anyone else who wants to play, the L2 cache control register is at physical address 0x120000078. The low 16 bits are the relevant ones. By default the SRM sets it to 0x1141. Other useful values are: 0x1140 - Disables L2 cache. Note that this will probably make the SRM reboot since it's a write-back cache. I found that the updated value made it through that reboot and everything else worked OK. 0x1155 - Enable L2 cache ECC and tag parity. (I have no idea why the SRM doesn't do this by default). 0x1255 - Enable L2 cache ECC and tag parity. Increase L2 reads by 1 cycle. BTW, you can get free samples of SN74ABT623DW from TI at http://www.ti.com/sc/docs/products/logic/sn74abt623.html#Pricing_Samples This is a drop-in replacement for the 74F623 and will run cooler. Peter To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011211154137.A73243>