Skip to contentSkip to conversation

Linux 2.6.6 "IDE cache-flush at shutdown fixes"

DISCUSSION Rene Herman 15 participants 34 messages May 10, 2004 – May 17, 2004 3 messages not in the archive

lore
Share & tools
lore
#0 lore raw

Good day.

The 2.6.6-rc3 -> 2.6.6-final changes to ide-disk.c unfortunately make my machine complain loudly both at boot and reboot:

Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
AMD7409: IDE controller at PCI slot 0000:00:07.1
AMD7409: chipset revision 7
AMD7409: not 100% native mode: will probe irqs later
AMD7409: 0000:00:07.1 (rev 07) UDMA66 controller

    ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda:DMA, hdb:pio
    ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:DMA, hdd:pio

hda: Maxtor 6Y120P0, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hdc: PLEXTOR DVD-ROM PX-116A, ATAPI CD/DVD-ROM drive
ide1 at 0x170-0x177,0x376 on irq 15
hda: max request size: 128KiB
hda: 240121728 sectors (122942 MB) w/7936KiB Cache, CHS=65535/16/63, UDMA(66)
hda: hda1 < hda5 hda6 hda7 hda8 hda9 hda10 hda11 hda12 hda13 > hda2
hda3 hda4
hda2: <bsd: hda14 hda15 hda16 hda17 hda18 >
hda4: <minix: hda19 hda20 >
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!

The disk, 6Y120P0, is a new-ish Maxtor "DiamondMax Plus 9", 120G, 8M cache. Controller is an AMD756. Same complaints on reboot. Reverting the rc3->final changes to ide-disk.c fixes/supresses them again.

Rene.

#1 lore raw
Quoted 13 lines · Rene Herman

Good day.

The 2.6.6-rc3 -> 2.6.6-final changes to ide-disk.c unfortunately make my machine complain loudly both at boot and reboot:

Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx AMD7409: IDE controller at PCI slot 0000:00:07.1
AMD7409: chipset revision 7
AMD7409: not 100% native mode: will probe irqs later
AMD7409: 0000:00:07.1 (rev 07) UDMA66 controller

    ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda:DMA, hdb:pio
    ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:DMA, hdd:pio
Quoted 15 lines

hda: Maxtor 6Y120P0, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hdc: PLEXTOR DVD-ROM PX-116A, ATAPI CD/DVD-ROM drive
ide1 at 0x170-0x177,0x376 on irq 15
hda: max request size: 128KiB
hda: 240121728 sectors (122942 MB) w/7936KiB Cache, CHS=65535/16/63, UDMA(66)
hda: hda1 < hda5 hda6 hda7 hda8 hda9 hda10 hda11 hda12 hda13 >
hda2 hda3 hda4
hda2: <bsd: hda14 hda15 hda16 hda17 hda18 >
hda4: <minix: hda19 hda20 >
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error } hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!

I have this too:
Linux version 2.6.6 ([email protected]) (gcc version 3.3.2 20031022 (Red Hat Linux 3.3.2-1)) #1 Mon May 10 02:20:54 EDT 2004 [...]
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
VP_IDE: IDE controller at PCI slot 0000:00:11.1
VP_IDE: chipset revision 6
VP_IDE: not 100% native mode: will probe irqs later
VP_IDE: VIA vt8233 (rev 00) IDE UDMA100 controller on pci0000:00:11.1

    ide0: BM-DMA at 0xd800-0xd807, BIOS settings: hda:DMA, hdb:DMA

[...]
hda: Maxtor 6Y120P0, ATA DISK drive
[...]
hda: max request size: 128KiB
hda: 240121728 sectors (122942 MB) w/7936KiB Cache, CHS=65535/16/63, UDMA(100)
hda: hda1 hda2 hda3 hda4 < hda5 hda6 hda7 hda8 >
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!

Quoted 5 lines

The disk, 6Y120P0, is a new-ish Maxtor "DiamondMax Plus 9", 120G, 8M cache. Controller is an AMD756. Same complaints on reboot. Reverting the rc3->final changes to ide-disk.c fixes/supresses them again.

Rene.

I note the drive is the same model here too, Rene.

The question remains however, is our data in danger?

Signature · 8 lines
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.22% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attorneys please note, additions to this message
by Gene Heskett are:
Copyright 2004 by Maurice Eugene Heskett, all rights reserved.
#2 lore raw
Quoted 5 lines · Gene Heskett
Quoted 1 line

hda: Maxtor 6Y120P0, ATA DISK drive

I note the drive is the same model here too, Rene.

The question remains however, is our data in danger?

There's a fair change that we'll be told, yes, very much so, since these drives don't seem to correctly support this life saving feature. The real answer though will be more easily deducted by calculating the ratio of unexplained file system corruptions you've had and reboots you've managed (0, that is).

Rene.

#3 lore raw
In reply to Gene Heskett
Quoted 4 lines

I note the drive is the same model here too, Rene.

The question remains however, is our data in danger?

You have a disk with an active write back cache but that doesn't support the flush-cache IDE command. That's obviously a bad combo ;) The other fixes in 2.6.6 ought to make it a bit less dangerous than before, eg the power off at shutdown WILL flush the cache.

#4 lore raw
In reply to Rene Herman

Hi,

Rene, can you test these (incremental) patches?

[PATCH] set drive->wcache only on the first ->open() call

drivers/ide/ide-disk.c 2004-05-07 18:30:09.544793448 +0200 renamed from drivers/ide/ide-disk.c.orig 2004-05-07 15:05:46.000000000 +0200 +2 -0
@@ -1758,6 +1758,8 @@
                if (drive->doorlocking && ide_raw_taskfile(drive, &args, NULL))
                        drive->doorlocking = 0;
        }
+       if (drive->usage != 1)
+               return 0;
        drive->wcache = 0;
        /* Cache enabled? */
        if (drive->id->csfo & 1)

Patch below reverts handling of flush cache to be _exactly_ the same as in 2.4 (no unknown commands on ->suspend() and ->shutdown()).

[PATCH] ide: don't send cacheflush to drives that don't understand it #2

drivers/ide/ide-disk.c 2004-05-07 18:34:18.766905928 +0200 renamed from drivers/ide/ide-disk.c.orig 2004-05-07 18:30:09.000000000 +0200 +1 -1
@@ -1758,7 +1758,7 @@
                if (drive->doorlocking && ide_raw_taskfile(drive, &args, NULL))
                        drive->doorlocking = 0;
        }
-       if (drive->usage != 1)
+       if (drive->usage != 1 || !drive->removable)
                return 0;
        drive->wcache = 0;
        /* Cache enabled? */
Quoted 4 lines · Rene Herman

Good day.

The 2.6.6-rc3 -> 2.6.6-final changes to ide-disk.c unfortunately make my machine complain loudly both at boot and reboot:

These warnings are _harmless_.

Your drives have write cache support
but don't understand flush cache commands.

Quoted 28 lines

Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
AMD7409: IDE controller at PCI slot 0000:00:07.1
AMD7409: chipset revision 7
AMD7409: not 100% native mode: will probe irqs later
AMD7409: 0000:00:07.1 (rev 07) UDMA66 controller

     ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda:DMA, hdb:pio
     ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:DMA, hdd:pio

hda: Maxtor 6Y120P0, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hdc: PLEXTOR DVD-ROM PX-116A, ATAPI CD/DVD-ROM drive
ide1 at 0x170-0x177,0x376 on irq 15
hda: max request size: 128KiB
hda: 240121728 sectors (122942 MB) w/7936KiB Cache, CHS=65535/16/63, UDMA(66)
hda: hda1 < hda5 hda6 hda7 hda8 hda9 hda10 hda11 hda12 hda13 > hda2
hda3 hda4
hda2: <bsd: hda14 hda15 hda16 hda17 hda18 >
hda4: <minix: hda19 hda20 >
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!

The disk, 6Y120P0, is a new-ish Maxtor "DiamondMax Plus 9", 120G, 8M cache. Controller is an AMD756. Same complaints on reboot. Reverting the rc3->final changes to ide-disk.c fixes/supresses them again.

Rene.

#5 lore raw

drivers/ide/ide-disk.c 2004-05-07 18:30:09.544793448 +0200 line 1762

Quoted 4 lines · Bartlomiej Zolnierkiewicz

Rene, can you test these (incremental) patches?

drivers/ide/ide-disk.c 2004-05-07 18:30:09.544793448 +0200 renamed from drivers/ide/ide-disk.c.orig 2004-05-07 15:05:46.000000000 +0200 +2 -0
+       if (drive->usage != 1)
+               return 0;

Only this one does not make a change.

drivers/ide/ide-disk.c 2004-05-07 18:34:18.766905928 +0200 line 1761

Quoted 2 lines
drivers/ide/ide-disk.c 2004-05-07 18:34:18.766905928 +0200 renamed from drivers/ide/ide-disk.c.orig 2004-05-07 18:30:09.000000000 +0200 +1 -1
-       if (drive->usage != 1)
+       if (drive->usage != 1 || !drive->removable)

With this one, the cache flushing noise is no more, but still a problem unfortunately. With or without these patches, 2.6.6 powers down the drive during reboot. This is very annoying, seeing as how it immediately needs to spin up again for POST.

Thanks for now!

Rene.

#6 lore raw
Quoted 1 line · Rene Herman

With this one, the cache flushing noise is no more, but still a problem unfortunately. With or without these patches, 2.6.6 powers down the drive during reboot. This is very annoying, seeing as how it immediately needs to spin up again for POST.

Sorry, mistaken. This only happens _with_ your change. Without, 2.6.6 just complains.

Rene.

#7 lore raw
Quoted 11 lines · On Mon, 10 May 2004 23:59:32 +0200 Rene Herman <[email protected]> wrote:
Quoted 4 lines · Rene Herman

With this one, the cache flushing noise is no more, but still a problem unfortunately. With or without these patches, 2.6.6 powers down the drive during reboot. This is very annoying, seeing as how it immediately needs to spin up again for POST.

Sorry, mistaken. This only happens _with_ your change. Without, 2.6.6 just complains.

Rene.

I am having this "spin up" behaviour with plain 2.6.6 (I haven't tried any patches)

On reboot, after it suspended the drives, there is a roughly 15 second delay at post while the bios is initializing the hard disks. During this time the hard disk LED is lit.

Shut down and power up behaviour is normal, it's only on a reboot that the above occurs.

Hardware: Pentium4, Gigabyte motherboard, i845, ICH2 IDE controllers.
Disks: Two identical Maxtor 40 Gb 6E040L0 models. Primary master and primary slave. UDMA (100).

Note that I do hear a bit of a "cache flush" noise, but I assume that's normal because Windows XP makes that same noise when shutting down. It seems to be a normal hard disk sound.

I do not get any complaints like you have in the boot logs though. Everything is normal unless I reboot (and even that is just a trivial delay and probably not at all harmful)

Mike

#8 lore raw
In reply to Rene Herman
Quoted 10 lines · you

drivers/ide/ide-disk.c 2004-05-07 18:30:09.544793448 +0200 line 1762

Quoted 4 lines · Bartlomiej Zolnierkiewicz

Rene, can you test these (incremental) patches?

drivers/ide/ide-disk.c 2004-05-07 18:30:09.544793448 +0200 renamed from drivers/ide/ide-disk.c.orig 2004-05-07 15:05:46.000000000 +0200 +2 -0
+       if (drive->usage != 1)
+               return 0;

Only this one does not make a change.

Quoted 2 lines
drivers/ide/ide-disk.c 2004-05-07 18:34:18.766905928 +0200 renamed from drivers/ide/ide-disk.c.orig 2004-05-07 18:30:09.000000000 +0200 +1 -1
-       if (drive->usage != 1)
+       if (drive->usage != 1 || !drive->removable)

Thanks.

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify? I still would like to know why these drives don't accept flush cache commands (or it is a driver's bug?).

Quoted 4 lines

With this one, the cache flushing noise is no more, but still a problem unfortunately. With or without these patches, 2.6.6 powers down the drive during reboot. This is very annoying, seeing as how it immediately needs to spin up again for POST.

There is a problem with new 2.6 generic ->shutdown framework, it doesn't differentiate between reboot / halt and power_off. We may try to fix it or revert to 2.4 way of doing things if this is too big change for 2.6.

Thanks,
Bartlomiej

#9 lore raw
Quoted 5 lines · Bartlomiej Zolnierkiewicz

There is a problem with new 2.6 generic ->shutdown framework,
it doesn't differentiate between reboot / halt and power_off. We may try to fix it or revert to 2.4 way of doing things if this is too big change for 2.6.

It's a bit grubby, but we could easily add a fourth state to `system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and SYSTEM_HALT. That would be a quite simple change.

#10 lore raw
Quoted 4 lines · Andrew Morton

It's a bit grubby, but we could easily add a fourth state to
`system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and SYSTEM_HALT. That would be a quite simple change.

Like this. I checked all the SYSTEM_FOO users and none of them seem to care about the shutdown state at present. Easy.

 25-akpm/include/linux/kernel.h |   11 +++++++----
 25-akpm/init/main.c            |    3 ++-
 25-akpm/kernel/sys.c           |    8 ++++----
 3 files changed, 13 insertions(+), 9 deletions(-)
include/linux/kernel.h renamed from include/linux/kernel.h~system-state-splitup +7 -4
diff -puN include/linux/kernel.h~system-state-splitup include/linux/kernel.h
@@ -109,14 +109,17 @@ static inline void console_verbose(void)
 extern void bust_spinlocks(int yes);
 extern int oops_in_progress;		/* If set, an oops, panic(), BUG() or die() is in progress */
 extern int panic_on_oops;
-extern int system_state;		/* See values below */
 extern int tainted;
 extern const char *print_tainted(void);
 
 /* Values used for system_state */
-#define SYSTEM_BOOTING 0
-#define SYSTEM_RUNNING 1
-#define SYSTEM_SHUTDOWN 2
+extern enum system_states {
+	SYSTEM_BOOTING,
+	SYSTEM_RUNNING,
+	SYSTEM_HALT,
+	SYSTEM_POWER_OFF,
+	SYSTEM_RESTART,
+} system_state;
 
 #define TAINT_PROPRIETARY_MODULE	(1<<0)
 #define TAINT_FORCED_MODULE		(1<<1)
init/main.c renamed from init/main.c~system-state-splitup +2 -1
diff -puN init/main.c~system-state-splitup init/main.c
@@ -95,7 +95,8 @@ extern void prepare_namespace(void);
 extern void tc_init(void);
 #endif
 
-int system_state;	/* SYSTEM_BOOTING/RUNNING/SHUTDOWN */
+enum system_states system_state;
+EXPORT_SYMBOL(system_state);
 
 /*
  * Boot command-line arguments
kernel/sys.c renamed from kernel/sys.c~system-state-splitup +4 -4
diff -puN kernel/sys.c~system-state-splitup kernel/sys.c
@@ -451,7 +451,7 @@ asmlinkage long sys_reboot(int magic1, i
 	switch (cmd) {
 	case LINUX_REBOOT_CMD_RESTART:
 		notifier_call_chain(&reboot_notifier_list, SYS_RESTART, NULL);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_RESTART;
 		device_shutdown();
 		printk(KERN_EMERG "Restarting system.\n");
 		machine_restart(NULL);
@@ -467,7 +467,7 @@ asmlinkage long sys_reboot(int magic1, i
 
 	case LINUX_REBOOT_CMD_HALT:
 		notifier_call_chain(&reboot_notifier_list, SYS_HALT, NULL);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_HALT;
 		device_shutdown();
 		printk(KERN_EMERG "System halted.\n");
 		machine_halt();
@@ -477,7 +477,7 @@ asmlinkage long sys_reboot(int magic1, i
 
 	case LINUX_REBOOT_CMD_POWER_OFF:
 		notifier_call_chain(&reboot_notifier_list, SYS_POWER_OFF, NULL);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_POWER_OFF;
 		device_shutdown();
 		printk(KERN_EMERG "Power down.\n");
 		machine_power_off();
@@ -493,7 +493,7 @@ asmlinkage long sys_reboot(int magic1, i
 		buffer[sizeof(buffer) - 1] = '\0';
 
 		notifier_call_chain(&reboot_notifier_list, SYS_RESTART, buffer);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_RESTART;
 		device_shutdown();
 		printk(KERN_EMERG "Restarting system with command '%s'.\n", buffer);
 		machine_restart(buffer);

_

#11 lore raw
Quoted 5 lines · Andrew Morton
Quoted 2 lines

It's a bit grubby, but we could easily add a fourth state to
`system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and SYSTEM_HALT. That would be a quite simple change.

Like this. I checked all the SYSTEM_FOO users and none of them seem to care about the shutdown state at present. Easy.

Wonderful. Placed the following quick hack on top:

[drivers/ide/ide-disk.c]

@@ -1704,10 +1704,11 @@

static void ide_device_shutdown(struct device *dev) {
- ide_drive_t *drive = container_of(dev, ide_drive_t, gendev);
-

-       printk("Shutdown: %s\n", drive->name);
-       dev->bus->suspend(dev, PM_SUSPEND_STANDBY);
+       if (system_state != SYSTEM_RESTART) {
+               ide_drive_t *drive = container_of(dev, ide_drive_t, gendev);
+               printk("Shutdown: %s\n", drive->name);
+               dev->bus->suspend(dev, PM_SUSPEND_STANDBY);

+ }
}

/*

Seems very wrong there; will likely want to be pushed up a few levels, but... Works For Me.

Have attached a patch of what I'm currently using against 2.6.6 just in case anyone interested lost track. It's bart+morton+hack.

Rene.

drivers/ide/ide-disk.c +7 -4
diff -urN linux-2.6.6.orig/drivers/ide/ide-disk.c linux-2.6.6/drivers/ide/ide-disk.c
@@ -1704,10 +1704,11 @@
 
 static void ide_device_shutdown(struct device *dev)
 {
-	ide_drive_t *drive = container_of(dev, ide_drive_t, gendev);
-
-	printk("Shutdown: %s\n", drive->name);
-	dev->bus->suspend(dev, PM_SUSPEND_STANDBY);
+	if (system_state != SYSTEM_RESTART) {
+		ide_drive_t *drive = container_of(dev, ide_drive_t, gendev);
+		printk("Shutdown: %s\n", drive->name);
+		dev->bus->suspend(dev, PM_SUSPEND_STANDBY);
+	}
 }
 
 /*
@@ -1758,6 +1759,8 @@
 		if (drive->doorlocking && ide_raw_taskfile(drive, &args, NULL))
 			drive->doorlocking = 0;
 	}
+	if (drive->usage != 1 || !drive->removable)
+		return 0;
 	drive->wcache = 0;
 	/* Cache enabled? */
 	if (drive->id->csfo & 1)
include/linux/kernel.h +7 -4
diff -urN linux-2.6.6.orig/include/linux/kernel.h linux-2.6.6/include/linux/kernel.h
@@ -109,14 +109,17 @@
 extern void bust_spinlocks(int yes);
 extern int oops_in_progress;		/* If set, an oops, panic(), BUG() or die() is in progress */
 extern int panic_on_oops;
-extern int system_state;		/* See values below */
 extern int tainted;
 extern const char *print_tainted(void);
 
 /* Values used for system_state */
-#define SYSTEM_BOOTING 0
-#define SYSTEM_RUNNING 1
-#define SYSTEM_SHUTDOWN 2
+extern enum system_states {
+	SYSTEM_BOOTING,
+	SYSTEM_RUNNING,
+	SYSTEM_HALT,
+	SYSTEM_POWER_OFF,
+	SYSTEM_RESTART,
+} system_state;
 
 #define TAINT_PROPRIETARY_MODULE	(1<<0)
 #define TAINT_FORCED_MODULE		(1<<1)
init/main.c +2 -1
diff -urN linux-2.6.6.orig/init/main.c linux-2.6.6/init/main.c
@@ -95,7 +95,8 @@
 extern void tc_init(void);
 #endif
 
-int system_state;	/* SYSTEM_BOOTING/RUNNING/SHUTDOWN */
+enum system_states system_state;
+EXPORT_SYMBOL(system_state);
 
 /*
  * Boot command-line arguments
kernel/sys.c +4 -4
diff -urN linux-2.6.6.orig/kernel/sys.c linux-2.6.6/kernel/sys.c
@@ -447,7 +447,7 @@
 	switch (cmd) {
 	case LINUX_REBOOT_CMD_RESTART:
 		notifier_call_chain(&reboot_notifier_list, SYS_RESTART, NULL);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_RESTART;
 		device_shutdown();
 		printk(KERN_EMERG "Restarting system.\n");
 		machine_restart(NULL);
@@ -463,7 +463,7 @@
 
 	case LINUX_REBOOT_CMD_HALT:
 		notifier_call_chain(&reboot_notifier_list, SYS_HALT, NULL);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_HALT;
 		device_shutdown();
 		printk(KERN_EMERG "System halted.\n");
 		machine_halt();
@@ -473,7 +473,7 @@
 
 	case LINUX_REBOOT_CMD_POWER_OFF:
 		notifier_call_chain(&reboot_notifier_list, SYS_POWER_OFF, NULL);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_POWER_OFF;
 		device_shutdown();
 		printk(KERN_EMERG "Power down.\n");
 		machine_power_off();
@@ -489,7 +489,7 @@
 		buffer[sizeof(buffer) - 1] = '\0';
 
 		notifier_call_chain(&reboot_notifier_list, SYS_RESTART, buffer);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_RESTART;
 		device_shutdown();
 		printk(KERN_EMERG "Restarting system with command '%s'.\n", buffer);
 		machine_restart(buffer);

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#14 lore raw
Quoted 48 lines · Rene Herman
Quoted 2 lines · Bartlomiej Zolnierkiewicz

Please revert ALL changes to 2.6.6 and gather some debug information using these simple patch.

Vanilla 2.6.6 with just this patch, at boot, directly after the partition scan:

hda: wcache=1 cmd=234
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!

At reboot or halt:

Shutdown: hda
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: DMA disabled
ide0: reset: success
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
ide0: reset: success
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError } Restarting system.

Quoted 2 lines

Oh and please check 'wcache' value from /proc/ide/hda/settings now and with this patch applied.

Vanilla 2.6.6:

rene@7ixe4:~/cache$ grep wcache settings-2.6.6
wcache                  1               0               1               rw
Quoted 8 lines

2.6.6 with the de{flush,spin}ification patches:

rene@7ixe4:~/cache$ grep wcache settings-2.6.6-hackedup
wcache                  0               0               1               rw

Hrmpf. 0?

Will test a few older Maxtor drives as well tonight. Hope it's useful.

Sorry for quoting everything, forgot to CC lkml first time.

Only took one. Does not happen on a (slightly) older Maxtor 4W030H2, 30G 5400RPM drive. 2.6.6 vanilla with your debug patch:

Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
PIIX4: IDE controller at PCI slot 0000:00:07.1
PIIX4: chipset revision 1
PIIX4: not 100% native mode: will probe irqs later

    ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda:DMA, hdb:DMA
    ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:DMA, hdd:DMA

hda: Maxtor 4W030H2, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hdc: ATAPI-CD ROM-DRIVE-52MAX, ATAPI CD/DVD-ROM drive
hdd: 6X4X32, ATAPI CD/DVD-ROM drive
ide1 at 0x170-0x177,0x376 on irq 15
hda: max request size: 128KiB
hda: 60030432 sectors (30735 MB) w/2048KiB Cache, CHS=59554/16/63, UDMA(33)
hda: hda1 hda2 hda3 hda4
hdc: ATAPI 52X CD-ROM drive, 128kB Cache, UDMA(33)
Uniform CD-ROM driver Revision: 3.20
hdd: ATAPI 32X CD-ROM CD-R/RW drive, 2048kB Cache, DMA

rene@5bt0:~$ su -c "cat /proc/ide/hda/settings" | grep wcache
wcache                  1               0               1               rw

Rene.

A message that is not in the archive (sent off-list or not yet received): [email protected]

#16 lore raw
Quoted 5 lines · Bartlomiej Zolnierkiewicz
Quoted 5 lines

Vanilla 2.6.6:

rene@7ixe4:~/cache$ grep wcache settings-2.6.6
wcache                  1               0               1               rw
Quoted 8 lines
Quoted 6 lines

2.6.6 with the de{flush,spin}ification patches:

rene@7ixe4:~/cache$ grep wcache settings-2.6.6-hackedup
wcache                  0               0               1               rw

Hrmpf. 0?

Very interesting, can you try vanilla 2.6.5 (or some other 2.6.x)?

2.4.26 : wcache = 0, 0, 1
2.6.6-rc3-mm2 : wcache = 0, 0, 1
2.6.6 : wcache = 1, 0, 1
2.6.6-hackedup : wcache = 0, 0, 1

(on none of these hdparm -W0/1 has any effect, by the way)

Quoted 1 line

We are getting close to the real problem.

Indeed. I have attached two "tiobench" seq. write results, one for vanilla 2.6.6, and one for 2.6.6-hackedup (your changes). I remember the result for the latter are the normal ones I got with previous kernels as well. The differences are very interesting.

2.6.6-vanilla (ie, the one showing wcache=1): +/- 10MB/s
2.6.6-hackedup (ie, shows wcache=0) : +/- 46MB/s

As said, that 46 is the expected value. Is this a logic inversion bug in either driver (but then both 2.4.26 and 2.6) or drive firmware?

Rene.

Sequential Writes

                              File  Blk   Num                   Avg      Maximum      Lat%     Lat%    CPU
Identifier                    Size  Size  Thr   Rate  (CPU%)  Latency    Latency      >2s      >10s    Eff
---------------------------- ------ ----- ---  ------ ------ --------- -----------  -------- -------- -----
2.6.6                         1534  4096    1   10.74 9.440%     0.298     2641.08   0.00102  0.00000   114
2.6.6                         1534  4096    2   10.64 9.999%     0.586     7005.74   0.00331  0.00000   106
2.6.6                         1534  4096    4   10.64 9.730%     1.129    17964.34   0.02550  0.00025   109
2.6.6                         1534  4096    8   10.12 9.423%     2.099    28425.08   0.04065  0.00256   107
Sequential Writes
                              File  Blk   Num                   Avg      Maximum      Lat%     Lat%    CPU
Identifier                    Size  Size  Thr   Rate  (CPU%)  Latency    Latency      >2s      >10s    Eff
---------------------------- ------ ----- ---  ------ ------ --------- -----------  -------- -------- -----
2.6.6                         1534  4096    1   46.59 57.01%     0.069      598.21   0.00000  0.00000    82
2.6.6                         1534  4096    2   42.44 50.34%     0.155     1951.60   0.00000  0.00000    84
2.6.6                         1534  4096    4   44.65 52.23%     0.267     1336.76   0.00000  0.00000    85
2.6.6                         1534  4096    8   37.78 41.34%     0.604     6720.81   0.00307  0.00000    91
#17 lore raw
In reply to Rene Herman
Quoted 5 lines · On Tue, 11 May 2004 13:24:10 +0200 Rene Herman <[email protected]> wrote:

Have attached a patch of what I'm currently using against 2.6.6 just in case anyone interested lost track. It's bart+morton+hack.

Rene.

[linux-2.6.6_rollup.diff text/plain (3726 bytes)]

I applied your bart+morton+hack rollup earlier today and it seems to be doing the right thing by me. The only problem I was having, was the suspend on restart and bios initialization delay (~ 15 seconds) because of it. That niggle was bugging me a bit and I'm glad to have it corrected.

So thanks Rene, Andrew and Bartlomiej :-)

Mike

#18 lore raw
In reply to Rene Herman
Quoted 11 lines · Rene Herman
Quoted 6 lines · Andrew Morton
Quoted 2 lines

It's a bit grubby, but we could easily add a fourth state to
`system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and SYSTEM_HALT. That would be a quite simple change.

Like this. I checked all the SYSTEM_FOO users and none of them seem to care about the shutdown state at present. Easy.

Wonderful. Placed the following quick hack on top:

	[snip]
Quoted 5 lines

Seems very wrong there; will likely want to be pushed up a few levels, but... Works For Me.

Have attached a patch of what I'm currently using against 2.6.6 just in case anyone interested lost track. It's bart+morton+hack.

Rene.

For which patch I say thank you, it saves a lot of time for people who have been following but not applying each level of fix as it came up.

Signature · 3 lines
   -bill davidsen ([email protected])
"The secret to procrastination is to put things off until the
 last possible moment - but no longer"  -me
#19 lore raw
In reply to Andrew Morton

Hi!

Quoted 6 lines
Quoted 3 lines

It's a bit grubby, but we could easily add a fourth state to
`system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and SYSTEM_HALT. That would be a quite simple change.

Like this. I checked all the SYSTEM_FOO users and none of them seem to care about the shutdown state at present. Easy.

Perhaps this should be parameter to device_shutdown? This is quite ugly.

Pavel

Quoted 25 lines

diff -puN include/linux/kernel.h~system-state-splitup include/linux/kernel.h

include/linux/kernel.h renamed from include/linux/kernel.h~system-state-splitup +7 -4
@@ -109,14 +109,17 @@ static inline void console_verbose(void)
 extern void bust_spinlocks(int yes);
 extern int oops_in_progress;		/* If set, an oops, panic(), BUG() or die() is in progress */
 extern int panic_on_oops;
-extern int system_state;		/* See values below */
 extern int tainted;
 extern const char *print_tainted(void);
 
 /* Values used for system_state */
-#define SYSTEM_BOOTING 0
-#define SYSTEM_RUNNING 1
-#define SYSTEM_SHUTDOWN 2
+extern enum system_states {
+	SYSTEM_BOOTING,
+	SYSTEM_RUNNING,
+	SYSTEM_HALT,
+	SYSTEM_POWER_OFF,
+	SYSTEM_RESTART,
+} system_state;
 
 #define TAINT_PROPRIETARY_MODULE	(1<<0)
 #define TAINT_FORCED_MODULE		(1<<1)

...

Quoted 12 lines

diff -puN kernel/sys.c~system-state-splitup kernel/sys.c

kernel/sys.c renamed from kernel/sys.c~system-state-splitup +1 -1
@@ -451,7 +451,7 @@ asmlinkage long sys_reboot(int magic1, i
 	switch (cmd) {
 	case LINUX_REBOOT_CMD_RESTART:
 		notifier_call_chain(&reboot_notifier_list, SYS_RESTART, NULL);
-		system_state = SYSTEM_SHUTDOWN;
+		system_state = SYSTEM_RESTART;
 		device_shutdown();
 		printk(KERN_EMERG "Restarting system.\n");
 		machine_restart(NULL);
Signature · 1 line
When do you have heart between your knees?
#20 lore raw
Quoted 10 lines · Pavel Machek
Quoted 7 lines
Quoted 4 lines

It's a bit grubby, but we could easily add a fourth state to
`system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and SYSTEM_HALT. That would be a quite simple change.

Like this. I checked all the SYSTEM_FOO users and none of them seem to care about the shutdown state at present. Easy.

Perhaps this should be parameter to device_shutdown? This is quite ugly.

Rather than a parameter to ->shutdown it would be better to add a new ->restart method to devices and IDE can implement one of those.

I don't know if it's worth the effort though. Is any other driver likely to want to discriminate between reboot and shutdown?

#21 lore raw
Quoted 8 lines · Andrew Morton
Quoted 7 lines · Pavel Machek
Quoted 6 lines
Quoted 3 lines

It's a bit grubby, but we could easily add a fourth state to
`system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and
SYSTEM_HALT. That would be a quite simple change.

Like this. I checked all the SYSTEM_FOO users and none of them seem to care about the shutdown state at present. Easy.

Quoted 8 lines
Quoted 2 lines

Perhaps this should be parameter to device_shutdown? This is quite ugly.

Rather than a parameter to ->shutdown it would be better to add a new ->restart method to devices and IDE can implement one of those.

I don't know if it's worth the effort though. Is any other driver likely to want to discriminate between reboot and shutdown?

it seems only drivers/char/watchdog/alim7101_wdt.c (currently uses reboot notifier for that)

#22 lore raw
In reply to Andrew Morton
Quoted 3 lines · Andrew Morton

I don't know if it's worth the effort though. Is any other driver likely to want to discriminate between reboot and shutdown?

e100 used to (and still does in 2.4) send the device into D3 on shutdown. This causes problems on a number of boards if the box is only rebooting as the driver fails to bring the device back out of D3.

Signature · 4 lines
Visit Openswan at http://www.openswan.org/
Email:  Herbert Xu ~{PmV>HI~} <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
#23 lore raw
Quoted 9 lines · Herbert Xu
Quoted 3 lines · Andrew Morton

I don't know if it's worth the effort though. Is any other driver likely to want to discriminate between reboot and shutdown?

e100 used to (and still does in 2.4) send the device into D3 on shutdown. This causes problems on a number of boards if the box is only rebooting as the driver fails to bring the device back out of D3.

Ho hum. Greg, any preferences? We can either:

a) Add a `restart' driver method and call that during reboot instead of ->shutdown, if the driver implements ->restart. Otherwise call ->shutdown or

b) stick with the

	if (system_state == SYSTEM_RESTART)
		...

thing in IDE and potentially a couple of other places?

#24 lore raw
Quoted 17 lines · Andrew Morton
Quoted 9 lines · Herbert Xu
Quoted 3 lines · Andrew Morton

I don't know if it's worth the effort though. Is any other driver likely to want to discriminate between reboot and shutdown?

e100 used to (and still does in 2.4) send the device into D3 on shutdown. This causes problems on a number of boards if the box is only rebooting as the driver fails to bring the device back out of D3.

Ho hum. Greg, any preferences? We can either:

a) Add a `restart' driver method and call that during reboot instead of ->shutdown, if the driver implements ->restart. Otherwise call ->shutdown or

Quoted 6 lines

b) stick with the

	if (system_state == SYSTEM_RESTART)
		...

thing in IDE and potentially a couple of other places?

I think we should stick with option b) for now, as we are already keeping this system state, right? The number of different drivers that will care about this is probably quite small.

But if I'm proven wrong, we can add "restart" to 2.7 :)

thanks,

greg k-h

#25 lore raw
Quoted 9 lines · Greg KH
Quoted 9 lines

Ho hum. Greg, any preferences? We can either:

a) Add a `restart' driver method and call that during reboot instead of ->shutdown, if the driver implements ->restart. Otherwise call ->shutdown or

b) stick with the

Quoted 7 lines
Quoted 4 lines
	if (system_state == SYSTEM_RESTART)
		...

thing in IDE and potentially a couple of other places?

I think we should stick with option b) for now, as we are already keeping this system state, right?

Yes, it's currently used in arch code which is otuside the driver model.

And the "current state of the entire system" maps comfortably onto a global variable.

#26 lore raw
In reply to Pavel Machek
Quoted 10 lines · Pavel Machek
Quoted 7 lines
Quoted 4 lines

It's a bit grubby, but we could easily add a fourth state to
`system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and SYSTEM_HALT. That would be a quite simple change.

Like this. I checked all the SYSTEM_FOO users and none of them seem to care about the shutdown state at present. Easy.

Perhaps this should be parameter to device_shutdown? This is quite ugly.

That was judged to be a 2.7 exercise. Seems to affect 143 files already.

#27 lore raw
In reply to Andrew Morton

Hi!

Quoted 8 lines
Quoted 4 lines

There is a problem with new 2.6 generic ->shutdown framework,
it doesn't differentiate between reboot / halt and power_off. We may try to fix it or revert to 2.4 way of doing things if this is too big change for 2.6.

It's a bit grubby, but we could easily add a fourth state to `system_state': split SYSTEM_SHUTDOWN into SYSTEM_REBOOT and SYSTEM_HALT. That would be a quite simple change.

I believe that we do not want to split that. These paths get pretty little testing, and splitting testing effort even more could be pretty bad.
Pavel

Signature · 1 line
When do you have heart between your knees?
#28 lore raw
In reply to Bartlomiej Zolnierkiewicz
Quoted 2 lines · Bartlomiej Zolnierkiewicz

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify?

Sure, attached. Quite sure you wanted hdc though? That's a DVD-ROM.

Quoted 1 line

I still would like to know why these drives don't accept flush cache commands (or it is a driver's bug?).

No idea I'm afraid. Seems at least new Maxtor drives are affected. Both the "120P0" (120G, 8M cache) and "L0" (120G, 2M cache) were reported in this thread.

Quoted 4 lines

There is a problem with new 2.6 generic ->shutdown framework, it doesn't differentiate between reboot / halt and power_off. We may try to fix it or revert to 2.4 way of doing things if this is too big change for 2.6.

Please also see reply to Andrew...

Rene.

# Maxtor 6Y120P0 (DiamondMax Plus 9, 120G, 8M cache)

0040 3fff c837 0010 0000 0000 003f 0000 0000 0000 5933 3232 4144 4a45 2020 2020 2020 2020 2020 2020 0003 3e00 0039 5941 5234 3142 5730 4d61 7874 6f72 2036 5931 3230 5030 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 8010 0000 2f00 4000 0200 0000 0007 0fcf 0010 00ff f310 00fb 0108 f780 0e4f 0000 0007 0003 0078 0078 0078 0078 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 00fe 001e 7c6b 7b09 4003 7c69 3a01 4003 107f 0000 0000 0000 fffe 600d c0c0 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0001 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0001 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 98a5

# Plextor PX-116A DVD-ROM

85c0 0000 0000 0000 0000 0000 0000 0000 0000 0000 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 0000 0000 0000 312e 3030 2020 2020 504c 4558 544f 5220 4456 442d 524f 4d20 5058 2d31 3136 4120 2020 2020 2020 2020 2020 2020 2020 2020 0000 0000 0b00 0000 0400 0200 0006 0000 0000 0000 0000 0000 0000 0000 0000 0000 0007 0003 0078 0078 00b4 0078 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 003e 0000 4218 4000 4000 4218 0000 4000 101f 0000 0000 0000 0000 6000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000

#29 lore raw
Quoted 5 lines · Rene Herman
Quoted 2 lines · Bartlomiej Zolnierkiewicz

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify?

Quoted 8 lines

Sure, attached. Quite sure you wanted hdc though? That's a DVD-ROM.

Quoted 2 lines

I still would like to know why these drives don't accept flush cache commands (or it is a driver's bug?).

No idea I'm afraid. Seems at least new Maxtor drives are affected. Both the "120P0" (120G, 8M cache) and "L0" (120G, 2M cache) were reported in this thread.

At a guess the 80P0 drives will also be affected (80G, 8mb cache), but as yet I havent tried 2.6.6 on the boxes with them. Tonight if theres time.

Craig

#30 lore raw
Quoted 5 lines · Craig Bradney
Quoted 2 lines
Quoted 2 lines

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify?

At a guess the 80P0 drives will also be affected (80G, 8mb cache), but as yet I havent tried 2.6.6 on the boxes with them. Tonight if theres time.

/dev/hda:

Model=Maxtor 6Y080P0, FwRev=YAR41BW0, SerialNo=Y24X835E Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=57 BuffType=DualPortCache, BuffSize=7936kB, MaxMultSect=16, MultSect=off CurCHS=4047/16/255, CurSects=16511760, LBA=yes, LBAsects=160086528 IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120} PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5 udma6 AdvancedPM=yes: disabled (255) WriteCache=enabled
Drive conforms to: (null):

At boot-time I did get two lots of:

hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!

Still on the first boot of 2.6.6 atm, so don't know if it bitches at shutdown too.

16:50:02 0$ cat /proc/ide/hda/identify 0040 3fff c837 0010 0000 0000 003f 0000 0000 0000 5932 3458 3833 3545 2020 2020 2020 2020 2020 2020 0003 3e00 0039 5941 5234 3142 5730 4d61 7874 6f72 2036 5930 3830 5030 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 8010 0000 2f00 4000 0200 0000 0007 0fcf 0010 00ff f310 00fb 0100 ba00 098a 0000 0007 0003 0078 0078 0078 0078 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 00fe 001e 7c6b 7b09 4003 7c69 3a01 4003 207f 0000 0000 0000 fffe 600b c0fe 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0009 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0001 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 d6a5 root@emelia:~;
16:50:03 0$

-Ath

Signature · 4 lines
- Athanasius = Athanasius(at)miggy.org / http://www.miggy.org/
                  Finger athan(at)fysh.org for PGP key
	   "And it's me who is my enemy. Me who beats me up.
Me who makes the monsters. Me who strips my confidence." Paula Cole - ME
#31 lore raw
In reply to Craig Bradney
Quoted 24 lines · Craig Bradney
Quoted 12 lines · Rene Herman
Quoted 4 lines · Bartlomiej Zolnierkiewicz

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify?

Sure, attached. Quite sure you wanted hdc though? That's a DVD-ROM.

Quoted 8 lines
Quoted 3 lines

I still would like to know why these drives don't accept flush cache commands (or it is a driver's bug?).

No idea I'm afraid. Seems at least new Maxtor drives are affected. Both the "120P0" (120G, 8M cache) and "L0" (120G, 2M cache) were reported in this thread.

Quoted 6 lines

At a guess the 80P0 drives will also be affected (80G, 8mb cache), but as yet I havent tried 2.6.6 on the boxes with them. Tonight if theres time.

Craig

I have Maxtor 6Y060L0 and is also affected. Now I am with 2.6.5.
SvrWks IDE controller also have problems with 2.6.6 because the drive works in mdma2 mode. When in 2.6.5 the transfer mode is udma2.
Probably because of this (from patch-2.6.6):

drivers/ide/pci/serverworks.c Sun May 9 19:33:36 2004 +2 -1
diff -Nru a/drivers/ide/pci/serverworks.c b/drivers/ide/pci/serverworks.c
@@ -472,7 +472,9 @@
                               int dma = config_chipset_for_dma(drive);
                               if ((id->field_valid & 2) && !dma)
                                       goto try_dma_modes;
-                       }
+                       } else
+                               /* UDMA disabled by mask, try other DMA modes */+                               goto try_dma_modes;
               } else if (id->field_valid & 2) {

try_dma_modes:

                       if ((id->dma_mword & hwif->mwdma_mask) ||

Here is part of dmesg from 2.6.6:

ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx SvrWks OSB4: IDE controller at PCI slot 0000:00:0f.1
SvrWks OSB4: chipset revision 0
SvrWks OSB4: not 100% native mode: will probe irqs later
ide0: BM-DMA at 0xffa0-0xffa7, BIOS settings: hda:DMA, hdb:pio
ide1: BM-DMA at 0xffa8-0xffaf, BIOS settings: hdc:pio, hdd:pio
hda: Maxtor 6Y060L0, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hda: max request size: 128KiB
hda: 120103200 sectors (61492 MB) w/2048KiB Cache, CHS=65535/16/63, (U)DMA
hda: hda1 hda2 hda3 < hda5 hda6 hda7 >
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!
mice: PS/2 mouse device common for all mice
input: PC Speaker
serio: i8042 AUX port at 0x60,0x64 irq 12
input: GenPS/2 Genius Wheel Mouse on isa0060/serio1
serio: i8042 KBD port at 0x60,0x64 irq 1
input: AT Translated Set 2 keyboard on isa0060/serio0
NET: Registered protocol family 2
IP: routing cache hash table of 2048 buckets, 16Kbytes
TCP: Hash tables configured (established 16384 bind 16384)
NET: Registered protocol family 1
NET: Registered protocol family 17
BIOS EDD facility v0.13 2004-Mar-09, 1 devices found
Please report your BIOS at http://linux.dell.com/edd/results.html
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!
VFS: Mounted root (ext2 filesystem) readonly.
Freeing unused kernel memory: 368k freed
Adding 297192k swap on /dev/hda2. Priority:-1 extents:1
Linux agpgart interface v0.100 (c) Dave Jones

#32 lore raw
Quoted 15 lines · dobrev
Quoted 14 lines · Craig Bradney
Quoted 12 lines · Rene Herman
Quoted 2 lines · Bartlomiej Zolnierkiewicz

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify?

Sure, attached. Quite sure you wanted hdc though? That's a DVD-ROM.

Quoted 2 lines

I still would like to know why these drives don't accept flush cache commands (or it is a driver's bug?).

No idea I'm afraid. Seems at least new Maxtor drives are affected. Both the "120P0" (120G, 8M cache) and "L0" (120G, 2M cache) were reported in this thread.

Quoted 7 lines
Quoted 5 lines

At a guess the 80P0 drives will also be affected (80G, 8mb cache), but as yet I havent tried 2.6.6 on the boxes with them. Tonight if theres time.

Craig

I have Maxtor 6Y060L0 and is also affected. Now I am with 2.6.5.

Please see http://bugme.osdl.org/show_bug.cgi?id=2672

Quoted 3 lines

SvrWks IDE controller also have problems with 2.6.6 because the drive works in mdma2 mode.
When in 2.6.5 the transfer mode is udma2.

UDMA2 on OSB4? Weird.

from serverwoks.c:

	/* If we are about to put a disk into UDMA mode we screwed up.
	   Our code assumes we never _ever_ do this on an OSB4 */

	if(dev->device == PCI_DEVICE_ID_SERVERWORKS_OSB4 &&
		drive->media == ide_disk && speed >= XFER_UDMA_0)
			BUG();

I need more data: .config (2.6.5/2.6.6) and full dmesg output (2.6.5/2.6.6).

Quoted 53 lines

Probably because of this (from patch-2.6.6):
diff -Nru a/drivers/ide/pci/serverworks.c b/drivers/ide/pci/serverworks.c

drivers/ide/pci/serverworks.c Sun May 9 19:33:36 2004 +2 -1
@@ -472,7 +472,9 @@
                                int dma = config_chipset_for_dma(drive);
                                if ((id->field_valid & 2) && !dma)
                                        goto try_dma_modes;
-                       }
+                       } else
+                               /* UDMA disabled by mask, try other DMA
 modes */+                               goto try_dma_modes;
                } else if (id->field_valid & 2) {
 try_dma_modes:
                        if ((id->dma_mword & hwif->mwdma_mask) ||

Here is part of dmesg from 2.6.6:

ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx SvrWks OSB4: IDE controller at PCI slot 0000:00:0f.1
SvrWks OSB4: chipset revision 0
SvrWks OSB4: not 100% native mode: will probe irqs later

    ide0: BM-DMA at 0xffa0-0xffa7, BIOS settings: hda:DMA, hdb:pio
    ide1: BM-DMA at 0xffa8-0xffaf, BIOS settings: hdc:pio, hdd:pio

hda: Maxtor 6Y060L0, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hda: max request size: 128KiB
hda: 120103200 sectors (61492 MB) w/2048KiB Cache, CHS=65535/16/63, (U)DMA
hda: hda1 hda2 hda3 < hda5 hda6 hda7 >
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!
mice: PS/2 mouse device common for all mice
input: PC Speaker
serio: i8042 AUX port at 0x60,0x64 irq 12
input: GenPS/2 Genius Wheel Mouse on isa0060/serio1
serio: i8042 KBD port at 0x60,0x64 irq 1
input: AT Translated Set 2 keyboard on isa0060/serio0
NET: Registered protocol family 2
IP: routing cache hash table of 2048 buckets, 16Kbytes
TCP: Hash tables configured (established 16384 bind 16384)
NET: Registered protocol family 1
NET: Registered protocol family 17
BIOS EDD facility v0.13 2004-Mar-09, 1 devices found
Please report your BIOS at http://linux.dell.com/edd/results.html
hda: task_no_data_intr: status=0x51 { DriveReady SeekComplete Error }
hda: task_no_data_intr: error=0x04 { DriveStatusError }
hda: Write Cache FAILED Flushing!
VFS: Mounted root (ext2 filesystem) readonly.
Freeing unused kernel memory: 368k freed
Adding 297192k swap on /dev/hda2. Priority:-1 extents:1
Linux agpgart interface v0.100 (c) Dave Jones

#33 lore raw
Quoted 33 lines · Bartlomiej Zolnierkiewicz
Quoted 30 lines · dobrev
Quoted 27 lines · Craig Bradney
Quoted 11 lines · Rene Herman
Quoted 4 lines · Bartlomiej Zolnierkiewicz

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify?

Sure, attached. Quite sure you wanted hdc though? That's a DVD-ROM.

Quoted 9 lines
Quoted 4 lines

I still would like to know why these drives don't accept flush cache commands (or it is a driver's bug?).

No idea I'm afraid. Seems at least new Maxtor drives are affected. Both the "120P0" (120G, 8M cache) and "L0" (120G, 2M cache) were reported in this thread.

At a guess the 80P0 drives will also be affected (80G, 8mb cache), but as yet I havent tried 2.6.6 on the boxes with them. Tonight if theres time.

Quoted 10 lines
Quoted 6 lines
Quoted 3 lines

Craig

I have Maxtor 6Y060L0 and is also affected. Now I am with 2.6.5.

Please see http://bugme.osdl.org/show_bug.cgi?id=2672

Yes, I know.

Quoted 12 lines
Quoted 7 lines

SvrWks IDE controller also have problems with 2.6.6 because the drive works in mdma2 mode.
When in 2.6.5 the transfer mode is udma2.

UDMA2 on OSB4? Weird.

from serverwoks.c:

Quoted 10 lines
	/* If we are about to put a disk into UDMA mode we screwed up.
	   Our code assumes we never _ever_ do this on an OSB4 */

	if(dev->device == PCI_DEVICE_ID_SERVERWORKS_OSB4 &&
		drive->media == ide_disk && speed >= XFER_UDMA_0)
			BUG();

I need more data: .config (2.6.5/2.6.6) and full dmesg output (2.6.5/2.6.6).

Yes, it's a OSB4.
I attached files you need. .config is the same.
The problem is that when in 2.6.6 hdparm reports that the drive is much slower than 2.6.5 2.6.6 => 13 MB/s
2.6.5 => 23 MB/s
When I remove the code related to serverworks.c (see bellow) in patch-2.6.6 transfer is like 2.6.5.

Quoted 19 lines
Quoted 19 lines

Probably because of this (from patch-2.6.6):
diff -Nru a/drivers/ide/pci/serverworks.c b/drivers/ide/pci/serverworks.c

drivers/ide/pci/serverworks.c Sun May 9 19:33:36 2004 +2 -1
@@ -472,7 +472,9 @@
                               int dma = config_chipset_for_dma(drive);
                               if ((id->field_valid & 2) && !dma)
                                       goto try_dma_modes;
-                       }
+                       } else
+                               /* UDMA disabled by mask, try other DMA
 modes */+                               goto try_dma_modes;
               } else if (id->field_valid & 2) {

try_dma_modes:

                       if ((id->dma_mword & hwif->mwdma_mask) ||
#34 lore raw
Quoted 73 lines · dobrev
Quoted 33 lines · Bartlomiej Zolnierkiewicz
Quoted 30 lines · dobrev
Quoted 27 lines · Craig Bradney
Quoted 11 lines · Rene Herman
Quoted 4 lines · Bartlomiej Zolnierkiewicz

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify?

Sure, attached. Quite sure you wanted hdc though? That's a DVD-ROM.

Quoted 9 lines
Quoted 4 lines

I still would like to know why these drives don't accept flush cache commands (or it is a driver's bug?).

No idea I'm afraid. Seems at least new Maxtor drives are affected. Both the "120P0" (120G, 8M cache) and "L0" (120G, 2M cache) were reported in this thread.

At a guess the 80P0 drives will also be affected (80G, 8mb cache), but as yet I havent tried 2.6.6 on the boxes with them. Tonight if theres time.

Quoted 10 lines
Quoted 6 lines
Quoted 3 lines

Craig

I have Maxtor 6Y060L0 and is also affected. Now I am with 2.6.5.

Please see http://bugme.osdl.org/show_bug.cgi?id=2672

Yes, I know.

Quoted 12 lines
Quoted 7 lines

SvrWks IDE controller also have problems with 2.6.6 because the drive works in mdma2 mode.
When in 2.6.5 the transfer mode is udma2.

UDMA2 on OSB4? Weird.

from serverwoks.c:

Quoted 10 lines
	/* If we are about to put a disk into UDMA mode we screwed up.
	   Our code assumes we never _ever_ do this on an OSB4 */

	if(dev->device == PCI_DEVICE_ID_SERVERWORKS_OSB4 &&
		drive->media == ide_disk && speed >= XFER_UDMA_0)
			BUG();

I need more data: .config (2.6.5/2.6.6) and full dmesg output (2.6.5/2.6.6).

Quoted 8 lines

Yes, it's a OSB4.
I attached files you need. .config is the same.
The problem is that when in 2.6.6 hdparm reports that the drive is much slower than 2.6.5
2.6.6 => 13 MB/s
2.6.5 => 23 MB/s
When I remove the code related to serverworks.c (see bellow) in patch-2.6.6 transfer is like 2.6.5.

I believe what happens, is that with the old logic UDMA disks on OSB4 "fell through the cracks" in svwks_config_drive_xfer_rate(). It was basically a noop (unintentionally) and the settings were left at whatever BIOS set them to.
Looking through some old threads, it looks like UDMA was considered not safe on an OSB4.

Patrick

Quoted 24 lines
Quoted 19 lines
Quoted 19 lines

Probably because of this (from patch-2.6.6):
diff -Nru a/drivers/ide/pci/serverworks.c b/drivers/ide/pci/serverworks.c

drivers/ide/pci/serverworks.c Sun May 9 19:33:36 2004 +2 -1
@@ -472,7 +472,9 @@
                               int dma = config_chipset_for_dma(drive);
                               if ((id->field_valid & 2) && !dma)
                                       goto try_dma_modes;
-                       }
+                       } else
+                               /* UDMA disabled by mask, try other DMA
 modes */+                               goto try_dma_modes;
               } else if (id->field_valid & 2) {

try_dma_modes:

                       if ((id->dma_mword & hwif->mwdma_mask) ||
#35 lore raw
Quoted 115 lines · Patrick Wildi
Quoted 100 lines · dobrev
Quoted 54 lines · Bartlomiej Zolnierkiewicz
Quoted 49 lines · dobrev
Quoted 44 lines · Craig Bradney
Quoted 23 lines · Rene Herman
Quoted 6 lines · Bartlomiej Zolnierkiewicz

Rene, can you send me copies of /proc/ide/hda/identify and /proc/ide/hdc/identify?

Sure, attached. Quite sure you wanted hdc though? That's a DVD-ROM.

Quoted 6 lines

I still would like to know why these drives don't accept flush cache commands (or it is a driver's bug?).

Quoted 7 lines

No idea I'm afraid. Seems at least new Maxtor drives are affected. Both the "120P0" (120G, 8M cache) and "L0" (120G, 2M cache) were reported in this thread.

At a guess the 80P0 drives will also be affected (80G, 8mb cache), but as yet I havent tried 2.6.6 on the boxes with them. Tonight if theres time.

Craig

Quoted 10 lines
Quoted 5 lines

I have Maxtor 6Y060L0 and is also affected. Now I am with 2.6.5.

Please see http://bugme.osdl.org/show_bug.cgi?id=2672

Yes, I know.

Quoted 16 lines
Quoted 9 lines

SvrWks IDE controller also have problems with 2.6.6 because the drive works in mdma2 mode.
When in 2.6.5 the transfer mode is udma2.

UDMA2 on OSB4? Weird.

from serverwoks.c:

Quoted 12 lines
	/* If we are about to put a disk into UDMA mode we screwed up.
	   Our code assumes we never _ever_ do this on an OSB4 */

	if(dev->device == PCI_DEVICE_ID_SERVERWORKS_OSB4 &&
		drive->media == ide_disk && speed >= XFER_UDMA_0)
			BUG();

I need more data: .config (2.6.5/2.6.6) and full dmesg output (2.6.5/2.6.6).

Quoted 10 lines

Yes, it's a OSB4.
I attached files you need. .config is the same.
The problem is that when in 2.6.6 hdparm reports that the drive is much slower than 2.6.5
2.6.6 => 13 MB/s
2.6.5 => 23 MB/s
When I remove the code related to serverworks.c (see bellow) in patch-2.6.6 transfer is like 2.6.5.

Quoted 10 lines

I believe what happens, is that with the old logic UDMA disks on OSB4 "fell through the cracks" in svwks_config_drive_xfer_rate(). It was basically a noop (unintentionally) and the settings were left at whatever BIOS set them to.
Looking through some old threads, it looks like UDMA was considered not safe on an OSB4.

Patrick

I agree, I think config_chipset_for_dma() was never entered before 2.6.6. Transfer rate was udma2 and there were no problems. Same was with 2.4. I don't know why OSB4 was concidered not to work in udma.
I have no problems for a long time and I want to use udma2.

Quoted 31 lines
Quoted 28 lines
Quoted 23 lines
Quoted 23 lines

Probably because of this (from patch-2.6.6):
diff -Nru a/drivers/ide/pci/serverworks.c b/drivers/ide/pci/serverworks.c

drivers/ide/pci/serverworks.c Sun May 9 19:33:36 2004 +2 -1
@@ -472,7 +472,9 @@
                              int dma = config_chipset_for_dma(drive);
                              if ((id->field_valid & 2) && !dma)
                                      goto try_dma_modes;
-                       }
+                       } else
+                               /* UDMA disabled by mask, try other DMA
 modes */+                               goto try_dma_modes;
              } else if (id->field_valid & 2) {

try_dma_modes:

                      if ((id->dma_mword & hwif->mwdma_mask) ||
#36 lore raw

scsi shutdown flush, journaled fses

In reply to Rene Herman

so ide drives caching writes is ok now, but what about scsi drives? do reiserfs and ext3 both use proper write barriers (especially for data=ordered)?

Signature · 2 lines
Tom Vier <[email protected]>
DSA Key ID 0x15741ECE