From mboxrd@z Thu Jan 1 00:00:00 1970 From: balbi@ti.com (Felipe Balbi) Date: Thu, 31 Jul 2014 18:04:14 -0500 Subject: Bad Page dump (help) In-Reply-To: <20140731153027.GG19512@saruman.home> References: <20140729190700.GN17808@saruman.home> <20140729195547.GE30282@n2100.arm.linux.org.uk> <20140729201750.GO17808@saruman.home> <20140731145354.GM30282@n2100.arm.linux.org.uk> <20140731153027.GG19512@saruman.home> Message-ID: <20140731230414.GA8920@saruman.home> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hi, On Thu, Jul 31, 2014 at 10:30:27AM -0500, Felipe Balbi wrote: > > I'm suspecting that you either have badly a routed data bus connection > > on bits 23:16, or you have a bad RAM chip which corrupts those bits. > > > > It may be worth trying a simple assembly-level RAM test which pokes > > all locations with alternating 0xffffffff, 0x00000000 and verifies > > them. By that I mean odd word addresses set to one value, even set > > to the other, verify, reverse them, try again. > > > > Alternatively, if there's a proper ARM memory checker around, I think > > it would be well worth running it on this board to check the integrity > > of the hardware. > > I'll try u-boot's memtester and see what results it yields. alright, stock u-boot's memtest didn't help much, so I modified it a bit and started seen issues when I was testing only the first 4MiB of RAM. Here's one small snippet of the output: RET ADDR WR RD FAIL @ 80000010 exp fffffffd got ffeafffd FAIL @ 80001c30 exp fffffe3b got fffbfe3b FAIL @ 80000630 exp ffffff9b got fffbff9b FAIL @ 80000430 exp ffffffbb got ff5bffbb So what I just did is that I changed u-boot so it was test 768MiB and I'll leave this running for the next 4 days or so, then I'll report here. In any case, thanks for pointing me in the right direction. cheers -- balbi -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 819 bytes Desc: Digital signature URL: