From mboxrd@z Thu Jan 1 00:00:00 1970 From: Lin Ming Subject: Re: [git patches] libata updates for 3.3 Date: Tue, 17 Jan 2012 13:16:33 +0800 Message-ID: <1326777393.13517.39.camel@minggr> References: <20120109003255.GA6598@havoc.gtf.org> <4F12E57E.3090805@garzik.org> <1326676534.13517.3.camel@minggr> <1326691403.13517.21.camel@minggr> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: Received: from mga09.intel.com ([134.134.136.24]:10799 "EHLO mga09.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750899Ab2AQFQf (ORCPT ); Tue, 17 Jan 2012 00:16:35 -0500 In-Reply-To: Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Linus Torvalds Cc: Jeff Garzik , Andrew Morton , linux-ide@vger.kernel.org, LKML On Mon, 2012-01-16 at 11:55 -0800, Linus Torvalds wrote: > On Mon, Jan 16, 2012 at 11:29 AM, Linus Torvalds > wrote: > > > > and I do think that resume used to be faster in v3.2 than it is even > > with your patch. But maybe that's some rose-colored glasses. > > > > I'll recompile an old kernel to check. > > Confirmed. > > Plain v3.2 really *is* faster than current git, even current git with > your patch. > > In v3.2 I get this: > > [ 92.035600] ata_piix 0000:00:1f.2: PCI INT B -> GSI 19 (level, > low) -> IRQ 19 > [ 92.035610] ata_piix 0000:00:1f.2: setting latency timer to 64 > [ 92.036213] sd 0:0:0:0: [sda] Starting disk > [ 93.060471] ata2.00: failed to resume link (SControl 0) > [ 93.379963] ata1.01: failed to resume link (SControl 0) > [ 93.535802] ata1.00: SATA link up 3.0 Gbps (SStatus 123 SControl 300) > [ 93.535815] ata1.01: SATA link down (SStatus 0 SControl 0) > [ 93.543968] ata1.00: ACPI cmd ef/03:46:00:00:00:a0 (SET FEATURES) > filtered out > [ 93.544747] ata1.00: configured for UDMA/100 > [ 93.916453] PM: resume of devices complete after 1883.974 msecs > [ 93.916677] PM: Finishing wakeup. > [ 94.086800] ata2.01: failed to resume link (SControl 0) > [ 94.098408] ata2.00: SATA link down (SStatus 4 SControl 0) > [ 94.098429] ata2.01: SATA link down (SStatus 0 SControl 0) > > while current git (with your patch) gives me > > [ 108.115373] ata_piix 0000:00:1f.2: setting latency timer to 64 > [ 109.142010] ata2.00: failed to resume link (SControl 0) > [ 109.462004] ata1.01: failed to resume link (SControl 0) > [ 109.618065] ata1.00: SATA link up 3.0 Gbps (SStatus 123 SControl 300) > [ 109.618078] ata1.01: SATA link down (SStatus 0 SControl 0) > [ 109.626242] ata1.00: ACPI cmd ef/03:46:00:00:00:a0 (SET FEATURES) > filtered out > [ 109.627060] ata1.00: configured for UDMA/100 > [ 109.627240] sd 0:0:0:0: [sda] Starting disk > [ 110.170015] ata2.01: failed to resume link (SControl 0) > [ 110.181480] ata2.00: SATA link down (SStatus 4 SControl 0) > [ 110.181496] ata2.01: SATA link down (SStatus 0 SControl 0) > [ 110.182124] PM: resume of devices complete after 2066.945 msecs > [ 110.224533] PM: Finishing wakeup. > > so old kernels used to be a tiny bit faster despite not doing that > async thing (still slower than I'd like: I'd think that we should be No, old kernel did that async thing. Old kernel embedded the ata port suspend/resume code in host suspend/resume code and host has async suspend enabled. I split the ata port suspend/resume code out. ata: add ata port system PM callbacks http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=5ef4108 But ata port async suspend was not enabled in above patch. So I think this is a regression. Thanks, Lin Ming > able to resume devices in less than a second, but I don't know where > all the time goes) > > Linus