From bernard@blackham.com.au Sat Jul 23 05:46:47 2005 Return-Path: Received: from laptop.nigel.suspend2.net ([unix socket]) (authenticated user=nigel bits=0) by laptop.nigel.suspend2.net (Cyrus v2.1.16-Mandrake-RPM-2.1.16-5mdk) with LMTP; Sat, 23 Jul 2005 05:46:47 +1000 X-Sieve: CMU Sieve 2.2 Received: (qmail 4575 invoked by uid 89); 22 Jul 2005 19:46:46 -0000 Received: from unknown (HELO localhost) (127.0.0.1) by 0 with SMTP; 22 Jul 2005 19:46:46 -0000 Delivered-To: ncunningham:linuxmail.org@linuxmail.org Received: from pop24.pr.outblaze.com [205.158.62.125] by localhost with IMAP (fetchmail-6.2.5) for nigel@cunningham.myip.net.au (single-drop); Sat, 23 Jul 2005 05:46:46 +1000 (EST) Received: (qmail 9534 invoked by uid 0); 22 Jul 2005 19:43:13 -0000 X-OB-Received: from unknown (192.168.9.137) by mta45-1.us4.outblaze.com; 22 Jul 2005 19:43:13 -0000 Received: from usmail.cyclades.com (unknown [64.186.161.37]) by spf5-4.us4.outblaze.com (Postfix) with ESMTP id 4D291539CC for ; Fri, 22 Jul 2005 19:43:11 +0000 (GMT) Received: from mail.cyclades.com ([64.186.161.6]) by usmail.cyclades.com (Lotus Domino Release 6.5.1) with ESMTP id 2005072211392248-136806 ; Fri, 22 Jul 2005 11:39:22 -0700 Received: from localhost (localhost [127.0.0.1]) by mail.cyclades.com (Postfix) with ESMTP id 9D725917E for ; Fri, 22 Jul 2005 11:39:22 -0700 (PDT) Received: from mail.cyclades.com ([127.0.0.1]) by localhost (mail [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 08141-23 for ; Fri, 22 Jul 2005 11:39:21 -0700 (PDT) Received: from oracle.bridgewayconsulting.com.au (oracle.bridgewayconsulting.com.au [203.56.14.38]) by mail.cyclades.com (Postfix) with ESMTP id D00748BC0 for ; Fri, 22 Jul 2005 11:39:20 -0700 (PDT) Received: from amidala (dsl-202-72-159-76.wa.westnet.com.au [202.72.159.76]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by oracle.bridgewayconsulting.com.au (Postfix) with ESMTP id BF3C62E8004 for ; Sat, 23 Jul 2005 02:39:17 +0800 (WST) Received: by amidala (Postfix, from userid 1000) id EEEB21549A5; Sat, 23 Jul 2005 02:17:26 +0800 (WST) Date: Sat, 23 Jul 2005 02:17:26 +0800 From: Bernard Blackham To: Nigel Cunningham Subject: Re: New patchset. Message-ID: <20050722181726.GA8014@blackham.com.au> References: <1122010211.2367.1.camel@localhost> Mime-Version: 1.0 In-Reply-To: <1122010211.2367.1.camel@localhost> Organization: Dagobah Systems User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at cyclades.com X-MIMETrack: Itemize by SMTP Server on USMail/Cyclades(Release 6.5.1|January 21, 2004) at 07/22/2005 11:39:22, Serialize by Router on USMail/Cyclades(Release 6.5.1|January 21, 2004) at 07/22/2005 12:43:12, Serialize complete at 07/22/2005 12:43:12 Content-Disposition: inline Content-Type: text/plain; CHARSET=us-ascii X-Evolution-Source: imap://nigel@127.0.0.1/ Content-Transfer-Encoding: 8bit On Fri, Jul 22, 2005 at 03:30:11PM +1000, Nigel Cunningham wrote: > There's a new patchset scp'd to suspend2.net. Thanks! An assortment of compile fixes inlined. I get some very strange behaviour with this. Admittedly it's the first 2.6.13-rc kernel I've actually tried, but my machine will frequently freeze up for a second or two for the duration of suspending and resuming. Not just when doing disk I/O either - it happens right from preparing the page directory. It's as if interrupts are disabled and something spins for a while. Does anything leap to mind? On another interesting sidenote, I tried compiling suspend2ui_text with dietlibc - now whenever a message is sent to the kernel in response to a keystroke, the kernel dies quite horribly (recursive panics forever with no useful stack traces, PIDs, or EIPs). I don't know what dietlibc is smoking, but the kernel side of userui should be more robust than that. More digging to be done... Bernard. Index: linux/kernel/power/disk.c =================================================================== --- linux.orig/kernel/power/disk.c 2005-07-23 01:19:47.000000000 +0800 +++ linux/kernel/power/disk.c 2005-07-23 01:20:04.000000000 +0800 @@ -164,7 +164,7 @@ static void unprepare_processes(void) { platform_finish(); - thaw_processes(); + thaw_processes(FREEZER_ALL_THREADS); enable_nonboot_cpus(); pm_restore_console(); } Index: linux/drivers/net/irda/vlsi_ir.c =================================================================== --- linux.orig/drivers/net/irda/vlsi_ir.c 2005-07-23 01:19:47.000000000 +0800 +++ linux/drivers/net/irda/vlsi_ir.c 2005-07-23 01:20:04.000000000 +0800 @@ -1750,7 +1750,7 @@ vlsi_irda_dev_t *idev; pci_power_t new_state; - if (state.event < PM_EVENT_FREEZE || state.event > EVENT_SUSPEND ) { + if (state.event < PM_EVENT_FREEZE || state.event > PM_EVENT_SUSPEND ) { IRDA_ERROR("%s - %s: invalid pm state request: %u\n", __FUNCTION__, PCIDEV_NAME(pdev), state.event); return 0; Index: linux/drivers/net/wireless/airo.c =================================================================== --- linux.orig/drivers/net/wireless/airo.c 2005-07-23 01:19:47.000000000 +0800 +++ linux/drivers/net/wireless/airo.c 2005-07-23 01:20:04.000000000 +0800 @@ -2232,7 +2232,7 @@ u32 *vals = stats_rid.vals; clear_bit(JOB_STATS, &ai->flags); - if (ai->power) { + if (ai->power.event != PM_EVENT_ON) { up(&ai->sem); return; } @@ -2962,7 +2962,7 @@ break; } - if (ai->power || test_bit(FLAG_FLASHING, &ai->flags)) { + if (ai->power.event != PM_EVENT_ON || test_bit(FLAG_FLASHING, &ai->flags)) { up(&ai->sem); continue; } @@ -5496,7 +5496,7 @@ return -EAGAIN; disable_MAC(ai, 0); netif_device_detach(dev); - ai->power = state.event; + ai->power.event = state.event; cmd.cmd=HOSTSLEEP; issuecommand(ai, &cmd, &rsp); @@ -5515,7 +5515,7 @@ pci_restore_state(pdev); pci_enable_wake(pdev, pci_choose_state(pdev, ai->power), 0); - if (ai->power > 1) { + if (ai->power.event > PM_EVENT_FREEZE) { reset_card(dev, 0); mpi_init_descriptors(ai); setup_card(ai, dev->dev_addr, 0); @@ -7117,7 +7117,7 @@ int rc = 0; struct airo_info *ai = (struct airo_info *)dev->priv; - if (ai->power) + if (ai->power.event != PM_EVENT_ON) return 0; switch (cmd) { @@ -7196,7 +7196,7 @@ /* Get stats out of the card */ clear_bit(JOB_WSTATS, &local->flags); - if (local->power) { + if (local->power.event != PM_EVENT_ON) { up(&local->sem); return; }