From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: 2.6.19-rc6 : Spontaneous reboots, stack overflows - seems to implicate xfs, scsi, networking, SMP Date: Wed, 22 Nov 2006 20:10:13 -0800 (PST) Message-ID: <20061122.201013.112290046.davem@davemloft.net> References: <9a8748490611211551v2ebe88fel2bcf25af004c338a@mail.gmail.com> <9a8748490611220458w4d94d953v21f7a29a9f1bdb72@mail.gmail.com> <20061123011809.GY37654165@melbourne.sgi.com> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: jesper.juhl@gmail.com, chatz@melbourne.sgi.com, linux-kernel@vger.kernel.org, xfs@oss.sgi.com, xfs-masters@oss.sgi.com, netdev@vger.kernel.org, linux-scsi@vger.kernel.org Return-path: To: dgc@sgi.com In-Reply-To: <20061123011809.GY37654165@melbourne.sgi.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: netdev.vger.kernel.org From: David Chinner Date: Thu, 23 Nov 2006 12:18:09 +1100 > So, assuming the stacks less than 32 bytes are 32 bytes, we've got > 1380 bytes in the XFS stack there, On sparc64 just the XFS parts of the backtrace would be a minimum of 2816 bytes (each function has a minimum 8 * 16 byte stack frame, and there are about 22 calls in that trace). It's probably a lot more with local variables and such. It's way too much. You guys have to fix this stuff. If TCP's full send and receive path can be done in less function calls, XFS can allocate blocks in less too. I would even say 10 function calls deep to allocate file blocks is overkill, but 22 it just astronomically bad.