From mboxrd@z Thu Jan 1 00:00:00 1970 From: Peter Zijlstra Subject: Re: [Announce] 2.6.29-rc4-rt2 Date: Fri, 20 Feb 2009 08:49:11 +0100 Message-ID: <1235116151.4736.12.camel@laptop> References: <1235111568.3363.3.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: Thomas Gleixner , Thomas Gleixner , LKML , rt-users , Ingo Molnar , Steven Rostedt , Carsten Emde , Clark Williams , "jens.axboe" , Tejun Heo , FUJITA Tomonori , James Bottomley To: Thomas Meyer Return-path: Received: from casper.infradead.org ([85.118.1.10]:53755 "EHLO casper.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752681AbZBTHto (ORCPT ); Fri, 20 Feb 2009 02:49:44 -0500 In-Reply-To: <1235111568.3363.3.camel@localhost.localdomain> Sender: linux-rt-users-owner@vger.kernel.org List-ID: On Fri, 2009-02-20 at 07:32 +0100, Thomas Meyer wrote: > Are you interested in these BUG reports? Yes we are, thanks! > Without CONFIG_HIGHPTE kernel boots and seems to work correct, but I'm > hitting these two BUGs: They appear to be identical, so I'll consider them two instances of one bug :-) > [ 19.840100] BUG: sleeping function called from invalid context at arch/x86/mm/highmem_32.c:8 > [ 19.840105] in_atomic(): 0, irqs_disabled(): 1, pid: 718, name: IRQ-14 > [ 19.840110] Pid: 718, comm: IRQ-14 Tainted: G W 2.6.29-rc4-rt2-tip #55 > [ 19.840113] Call Trace: > [ 19.840123] [] __might_sleep+0xda/0xf0 > [ 19.840128] [] kmap+0x50/0x77 > [ 19.840133] [] sg_miter_next+0xc1/0xdf > [ 19.840137] [] sg_copy_buffer+0x95/0xc5 > [ 19.840141] [] sg_copy_to_buffer+0x1d/0x33 > [ 19.840149] [] atapi_qc_complete+0x25a/0x2c5 > [ 19.840153] [] __ata_qc_complete+0xb4/0xcd > [ 19.840157] [] ata_qc_complete+0x1be/0x1d7 > [ 19.840162] [] ata_hsm_qc_complete+0xa9/0xd3 > [ 19.840166] [] ata_sff_hsm_move+0x67b/0x6da > [ 19.840171] [] ata_sff_interrupt+0x15d/0x205 > [ 19.840176] [] handle_IRQ_event+0x4b/0xce > [ 19.840180] [] do_irqd+0x141/0x271 > [ 19.840184] [] ? do_irqd+0x0/0x271 > [ 19.840189] [] kthread+0x4a/0x82 > [ 19.840192] [] ? kthread+0x0/0x82 > [ 19.840197] [] kernel_thread_helper+0x7/0x10 sg_copy_buffer() seems to disable IRQs, however since on -rt everything is a irq-thread and kmap is preemptible, we don't appear to have that need. CC'd everybody who ever touched the file. --- lib/scatterlist.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/lib/scatterlist.c b/lib/scatterlist.c index b7b449d..4c6510d 100644 --- a/lib/scatterlist.c +++ b/lib/scatterlist.c @@ -426,7 +426,7 @@ static size_t sg_copy_buffer(struct scatterlist *sgl, unsigned int nents, sg_miter_start(&miter, sgl, nents, SG_MITER_ATOMIC); - local_irq_save(flags); + local_irq_save_nort(flags); while (sg_miter_next(&miter) && offset < buflen) { unsigned int len; @@ -445,7 +445,7 @@ static size_t sg_copy_buffer(struct scatterlist *sgl, unsigned int nents, sg_miter_stop(&miter); - local_irq_restore(flags); + local_irq_restore_nort(flags); return offset; }