/proc/partitions hosed in 2.4.9-ac10

/proc/partitions hosed in 2.4.9-ac10

[Erik Andersen]

[----------snip----------]
scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.1
        <Adaptec 2940 Ultra SCSI adapter>
        aic7880: Ultra Single Channel A, SCSI Id=7, 16/255 SCBs

(scsi0:A:4): 10.000MB/s transfers (10.000MHz, offset 15)
  Vendor: OLYMPUS   Model: MOS364            Rev: 1.02
  Type:   Optical Device                     ANSI SCSI revision: 02
(scsi0:A:5): 10.000MB/s transfers (10.000MHz, offset 15)
  Vendor: OLYMPUS   Model: MOS364            Rev: 1.02
  Type:   Optical Device                     ANSI SCSI revision: 02
[----------snip----------]
Attached scsi removable disk sda at scsi0, channel 0, id 4, lun 0
Attached scsi removable disk sdb at scsi0, channel 0, id 5, lun 0
SCSI device sda: 310352 2048-byte hdwr sectors (636 MB)
sda: Write Protect is off
 sda:<5>ll_rw_block: device 08:00: only 2048-char blocks implemented (1024)
 unable to read boot sectors / partition sectors
SCSI device sdb: 310352 2048-byte hdwr sectors (636 MB)
sdb: Write Protect is off
 sdb:<5>ll_rw_block: device 08:10: only 2048-char blocks implemented (1024)
 unable to read boot sectors / partition sectors
[----------snip----------]



$ cat /proc/partitions
major minor  #blocks  name

   8     0     620704 sda
   8    16     620704 sdb
   8     0     620704 sda
   8    16     620704 sdb
   8     0     620704 sda
   8    16     620704 sdb
   8     0     620704 sda
   8    16     620704 sdb
   8     0     620704 sda
   8    16     620704 sdb
   8     0     620704 sda
   8    16     620704 sdb
   8     0     620704 sda
   8    16     620704 sdb
   <continues forever>

In this case, there is no partition table on the magneto optical media
(since there seems little reason for such things)

    $ fdisk -l /dev/sda
    Note: sector size is 2048 (not 512)

    Disk /dev/sda: 64 heads, 32 sectors, 151 cylinders
    Units = cylinders of 2048 * 2048 bytes

    Disk /dev/sda doesn't contain a valid partition table
    $ 
    $ file -s /dev/sda
    /dev/sda: Linux rev 1.0 ext2 filesystem data

 -Erik

--
Erik B. Andersen   email:  andersee@debian.org, formerly of Lineo
--This message was written using 73% post-consumer electrons--

Re: /proc/partitions hosed in 2.4.9-ac10

[Erik Andersen]

On Mon Sep 17, 2001 at 03:19:57PM -0600, Erik Andersen wrote:
> $ cat /proc/partitions
> major minor  #blocks  name
> 
>    8     0     620704 sda
>    8    16     620704 sdb
>    8     0     620704 sda
>    8    16     620704 sdb
>    8     0     620704 sda
>    8    16     620704 sdb
>    8     0     620704 sda
>    8    16     620704 sdb
>    8     0     620704 sda
>    8    16     620704 sdb
>    8     0     620704 sda
>    8    16     620704 sdb
>    8     0     620704 sda
>    8    16     620704 sdb
>    <continues forever>
> 

Reverting drivers/scsi/sd.c to stock 2.4.9 fixes it, BTW,

 -Erik

--
Erik B. Andersen   email:  andersee@debian.org, formerly of Lineo
--This message was written using 73% post-consumer electrons--

Re: /proc/partitions hosed in 2.4.9-ac10

[Colonel]

In clouddancer.list.kernel, andersen@codepoet.org wrote:
>
>[----------snip----------]
>scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA DRIVER, Rev 6.2.1
>        <Adaptec 2940 Ultra SCSI adapter>
>        aic7880: Ultra Single Channel A, SCSI Id=7, 16/255 SCBs
...
>$ cat /proc/partitions
>major minor  #blocks  name
>
>   8     0     620704 sda
>   8    16     620704 sdb
>   8     0     620704 sda
>   8    16     620704 sdb
>   8     0     620704 sda
>   8    16     620704 sdb
>   8     0     620704 sda
>   8    16     620704 sdb
>   8     0     620704 sda
>   8    16     620704 sdb
>   8     0     620704 sda
>   8    16     620704 sdb
>   8     0     620704 sda
>   8    16     620704 sdb
>   <continues forever>


Works fine here:

Linux ns1.clouddancer.com 2.4.9-ac10 #1 SMP Tue Sep 11 21:47:15 PDT
2001 i686 unknown


 }cat /proc/partitions
major minor  #blocks  name

   9     0   53347584 md0
   8     0    8891620 sda
   8     1    4060160 sda1
   8     2    4403200 sda2
   8     3      34800 sda3
   8     4     393216 sda4
   8    16    8891620 sdb
   8    17    8891376 sdb1
   8    32    8891620 sdc
   8    33    8891376 sdc1
   8    48    8891620 sdd
   8    49    8891376 sdd1
   8    64    8891620 sde
   8    65    8891376 sde1
   8    80    8891620 sdf
   8    81    8891376 sdf1
   8    96    8891620 sdg
   8    97    8891376 sdg1
   8   112    8891620 sdh
   8   113    8891376 sdh1


SCSI subsystem driver Revision: 1.00
sym53c8xx: at PCI bus 0, device 9, function 0
sym53c8xx: setting PCI_COMMAND_PARITY...(fix-up)
sym53c8xx: 53c875 detected with Symbios NVRAM
sym53c8xx: at PCI bus 0, device 9, function 1


Perhaps it is a driver effect?

-- 
Windows 2001: "I'm sorry Dave ...  I'm afraid I can't do that."

Re: /proc/partitions hosed in 2.4.9-ac10

[Erik Andersen]

On Mon Sep 17, 2001 at 03:32:03PM -0700, Colonel wrote:
> 
> Works fine here:

But none of your devices have 2048 byte physical sectors, 
which is the case with my MO drives, and that appears to 
be the root of the problem,

 -Erik

--
Erik B. Andersen   email:  andersee@debian.org, formerly of Lineo
--This message was written using 73% post-consumer electrons--

Re: /proc/partitions hosed in 2.4.9-ac10

[Robert Macaulay]

I have the problem as well, and my physical sector size is 512. I'll try 
the patch

On Mon, 17 Sep 2001, Erik Andersen wrote:

> 
> On Mon Sep 17, 2001 at 03:32:03PM -0700, Colonel wrote:
> >
> > Works fine here:
> 
> But none of your devices have 2048 byte physical sectors,
> which is the case with my MO drives, and that appears to
> be the root of the problem,
> 
>  -Erik
> 
> --
> Erik B. Andersen   email:  andersee@debian.org, formerly of Lineo
> --This message was written using 73% post-consumer electrons--
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel"
> in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 
> 
> 

Re: /proc/partitions hosed in 2.4.9-ac10

[Hubert Mantel]

Hi,

On Mon, Sep 17, Erik Andersen wrote:

> On Mon Sep 17, 2001 at 03:32:03PM -0700, Colonel wrote:
> > 
> > Works fine here:
> 
> But none of your devices have 2048 byte physical sectors, 
> which is the case with my MO drives, and that appears to 
> be the root of the problem,

No, the problem only seems to exist when the low level SCSI drivers are 
being used as modules, not linked into the kernel.

Just tried it with 2.4.10-pre10: sym53c8xx and aic7xxx loaded as module
==> "cat /proc/partitions" gives infinite output (and my /dev/hda does
not show up at all). Both drivers compiled into the kernel ==> everything 
is fine; even my /dev/hda shows up.

>  -Erik
                                                                  -o)
    Hubert Mantel              Goodbye, dots...                   /\\
                                                                 _\_v

Re: /proc/partitions hosed in 2.4.9-ac10

[Hubert Mantel]

Hi,

On Mon, Sep 17, Colonel wrote:

> >$ cat /proc/partitions

[...]

> Works fine here:

[...]

> SCSI subsystem driver Revision: 1.00
> sym53c8xx: at PCI bus 0, device 9, function 0
> sym53c8xx: setting PCI_COMMAND_PARITY...(fix-up)
> sym53c8xx: 53c875 detected with Symbios NVRAM
> sym53c8xx: at PCI bus 0, device 9, function 1
> 
> 
> Perhaps it is a driver effect?

You only have one single SCSI adapter?

I tried several things so far, and it seems you need the following to 
trigger the problem: You need at least two SCSI adapters that require 
different drivers (so two AHA2940s are not sufficient) and the drivers 
need to be loaded as modules.
                                                                  -o)
    Hubert Mantel              Goodbye, dots...                   /\\
                                                                 _\_v

Re: /proc/partitions hosed in 2.4.9-ac10

[Bob McElrath]

[-- Attachment #1: Type: text/plain, Size: 1309 bytes --]

Hubert Mantel [mantel@suse.de] wrote:
> Hi,
> 
> On Mon, Sep 17, Colonel wrote:
> 
> > >$ cat /proc/partitions
> 
> [...]
> 
> > Works fine here:
> 
> [...]
> 
> > SCSI subsystem driver Revision: 1.00
> > sym53c8xx: at PCI bus 0, device 9, function 0
> > sym53c8xx: setting PCI_COMMAND_PARITY...(fix-up)
> > sym53c8xx: 53c875 detected with Symbios NVRAM
> > sym53c8xx: at PCI bus 0, device 9, function 1
> > 
> > 
> > Perhaps it is a driver effect?
> 
> You only have one single SCSI adapter?
> 
> I tried several things so far, and it seems you need the following to 
> trigger the problem: You need at least two SCSI adapters that require 
> different drivers (so two AHA2940s are not sufficient) and the drivers 
> need to be loaded as modules.

I have a BusLogic BT-958 (compiled in), pas16 (module), and usb-storage
(module).

The problem shows up with only the BusLogic scsi device, and the other
modules not loaded.  (And it also shows up with 2 or more scsi drivers
loaded)  Note that on this machine the pas16 and usb-storage modules
*were* installed, I removed them, and /proc/partitions still looped.  (I
can't readily reboot ATM)

Hope this helps,
-- Bob

Bob McElrath (rsmcelrath@students.wisc.edu) 
Univ. of Wisconsin at Madison, Department of Physics

[-- Attachment #2: Type: application/pgp-signature, Size: 240 bytes --]

Re: /proc/partitions hosed in 2.4.9-ac10

[Tim Walberg]

[-- Attachment #1: Type: text/plain, Size: 789 bytes --]

On 09/18/2001 17:43 +0200, Hubert Mantel wrote:
>>	
>>	You only have one single SCSI adapter?
>>	
>>	I tried several things so far, and it seems you need the following to 
>>	trigger the problem: You need at least two SCSI adapters that require 
>>	different drivers (so two AHA2940s are not sufficient) and the drivers 
>>	need to be loaded as modules.
>>	                                                                  -o)


I would amend that a bit - it doesn't seem to have to be only two SCSI
drivers, because I've seen the same with a 2.4.9-ac9 system with aic7xxx
(with sg, sd, and sr) combined with usb-storage (which also uses sd).
Granted usb-storage is kinda pseudo-SCSI, but it's not truly a SCSI
low level driver.

				tw


-- 
twalberg@mindspring.com

[-- Attachment #2: Type: application/pgp-signature, Size: 175 bytes --]

Re: /proc/partitions hosed in 2.4.9-ac10

[Thorsten Kranzkowski]

On Tue, Sep 18, 2001 at 10:59:32AM -0500, Tim Walberg wrote:
> On 09/18/2001 17:43 +0200, Hubert Mantel wrote:
> >>	
> >>	You only have one single SCSI adapter?
> >>	
> >>	I tried several things so far, and it seems you need the following to 
> >>	trigger the problem: You need at least two SCSI adapters that require 
> >>	different drivers (so two AHA2940s are not sufficient) and the drivers 
> >>	need to be loaded as modules.
> >>	                                                                  -o)
> 
> 
> I would amend that a bit - it doesn't seem to have to be only two SCSI
> drivers, because I've seen the same with a 2.4.9-ac9 system with aic7xxx
> (with sg, sd, and sr) combined with usb-storage (which also uses sd).
> Granted usb-storage is kinda pseudo-SCSI, but it's not truly a SCSI
> low level driver.
> 
> 				tw

As a datapoint: I have seen this when I "scsi remove-single-device"'d a disk
from my system. After re-adding it (at the same ID/LUN etc. in this case) 
/proc/partitions was OK again.

Two different Adaptec controllers driven by AIC7xxx, compiled into kernel,
no modules.

Thorsten


-- 
| Thorsten Kranzkowski        Internet: dl8bcu@dl8bcu.de                      |
| Mobile: ++49 170 1876134       Snail: Niemannsweg 30, 49201 Dissen, Germany |
| Ampr: dl8bcu@db0lj.#rpl.deu.eu, dl8bcu@marvin.dl8bcu.ampr.org [44.130.8.19] |

Re: /proc/partitions hosed in 2.4.9-ac10

[Christoph Hellwig]

In article <20010917151957.A26615@codepoet.org> you wrote:
> [----------snip----------]
> [----------snip----------]

Could you please try the attached patch agains 2.4.10-pre10?

--- ../master/linux-2.4.10-pre10/drivers/scsi/sd.c	Mon Sep 17 00:48:49 2001
+++ linux/drivers/scsi/sd.c	Tue Sep 18 00:34:41 2001
@@ -557,22 +557,7 @@
 	revalidate:		fop_revalidate_scsidisk
 };
 
-/*
- *    If we need more than one SCSI disk major (i.e. more than
- *      16 SCSI disks), we'll have to kmalloc() more gendisks later.
- */
-
-static struct gendisk sd_gendisk =
-{
-	major:		SCSI_DISK0_MAJOR,
-	major_name:	"sd",
-	minor_shift:	4,
-	max_p:		1 << 4,
-	fops:		&sd_fops,
-};
-
-static struct gendisk *sd_gendisks = &sd_gendisk;
-
+static struct gendisk *sd_gendisks;
 #define SD_GENDISK(i)    sd_gendisks[(i) / SCSI_DISKS_PER_MAJOR]
 
 /*
@@ -1132,33 +1117,34 @@
 		goto cleanup_sd;
 	memset(sd, 0, (sd_template.dev_max << 4) * sizeof(struct hd_struct));
 
-	if (N_USED_SD_MAJORS > 1)
-		sd_gendisks = kmalloc(N_USED_SD_MAJORS * sizeof(struct gendisk), GFP_ATOMIC);
-		if (!sd_gendisks)
-			goto cleanup_sd_gendisks;
+	sd_gendisks = kmalloc(N_USED_SD_MAJORS * sizeof(struct gendisk), GFP_ATOMIC);
+	if (!sd_gendisks)
+		goto cleanup_sd_gendisks;
+
 	for (i = 0; i < N_USED_SD_MAJORS; i++) {
-		sd_gendisks[i] = sd_gendisk;
-		sd_gendisks[i].de_arr = kmalloc (SCSI_DISKS_PER_MAJOR * sizeof *sd_gendisks[i].de_arr,
-                                                 GFP_ATOMIC);
-		if (!sd_gendisks[i].de_arr)
+		struct gendisk	*g = &sd_gendisks[i];
+
+		g->de_arr = kmalloc(SCSI_DISKS_PER_MAJOR * sizeof(*g->de_arr),
+				GFP_ATOMIC);
+		if (!g->de_arr)
 			goto cleanup_gendisks_de_arr;
-                memset (sd_gendisks[i].de_arr, 0,
-                        SCSI_DISKS_PER_MAJOR * sizeof *sd_gendisks[i].de_arr);
-		sd_gendisks[i].flags = kmalloc (SCSI_DISKS_PER_MAJOR * sizeof *sd_gendisks[i].flags,
-                                                GFP_ATOMIC);
-		if (!sd_gendisks[i].flags)
+		memset(g->de_arr, 0, SCSI_DISKS_PER_MAJOR * sizeof(*g->de_arr));
+
+		g->flags = kmalloc(SCSI_DISKS_PER_MAJOR *
+				sizeof(*g->flags), GFP_ATOMIC);
+		if (!g->flags)
 			goto cleanup_gendisks_flags;
-                memset (sd_gendisks[i].flags, 0,
-                        SCSI_DISKS_PER_MAJOR * sizeof *sd_gendisks[i].flags);
-		sd_gendisks[i].major = SD_MAJOR(i);
-		sd_gendisks[i].major_name = "sd";
-		sd_gendisks[i].minor_shift = 4;
-		sd_gendisks[i].max_p = 1 << 4;
-		sd_gendisks[i].part = sd + (i * SCSI_DISKS_PER_MAJOR << 4);
-		sd_gendisks[i].sizes = sd_sizes + (i * SCSI_DISKS_PER_MAJOR << 4);
-		sd_gendisks[i].nr_real = 0;
-		sd_gendisks[i].real_devices =
-		    (void *) (rscsi_disks + i * SCSI_DISKS_PER_MAJOR);
+		memset(g->flags, 0, SCSI_DISKS_PER_MAJOR * sizeof(*g->flags));
+
+		g->major = SD_MAJOR(i);
+		g->major_name = "sd";
+		g->minor_shift = 4;
+		g->max_p = 1 << g->minor_shift;
+		g->part = sd + (i * SCSI_DISKS_PER_MAJOR << 4);
+		g->sizes = sd_sizes + (i * SCSI_DISKS_PER_MAJOR << 4);
+		g->nr_real = 0;
+		g->real_devices = (rscsi_disks + i * SCSI_DISKS_PER_MAJOR);
+		g->fops = &sd_fops;
 	}
 
 	return 0;
@@ -1348,8 +1334,8 @@
 
 			/* If we are disconnecting a disk driver, sync and invalidate
 			 * everything */
-			max_p = sd_gendisk.max_p;
-			start = i << sd_gendisk.minor_shift;
+			max_p = sd_gendisks->max_p;
+			start = i << sd_gendisks->minor_shift;
 
 			for (j = max_p - 1; j >= 0; j--) {
 				int index = start + j;
@@ -1403,8 +1389,7 @@
 		read_ahead[SD_MAJOR(i)] = 0;
 	}
 	sd_template.dev_max = 0;
-	if (sd_gendisks != &sd_gendisk)
-		kfree(sd_gendisks);
+	kfree(sd_gendisks);
 }
 
 module_init(init_sd);

Re: /proc/partitions hosed in 2.4.9-ac10

[Hubert Mantel]

Hi,

On Tue, Sep 18, Christoph Hellwig wrote:

> Could you please try the attached patch agains 2.4.10-pre10?

Tried it. Does not help.
                                                                  -o)
    Hubert Mantel              Goodbye, dots...                   /\\
                                                                 _\_v

Re: /proc/partitions hosed in 2.4.9-ac10

[Christoph Hellwig]

On Tue, Sep 18, 2001 at 05:40:29PM +0200, Hubert Mantel wrote:
> Hi,
> 
> On Tue, Sep 18, Christoph Hellwig wrote:
> 
> > Could you please try the attached patch agains 2.4.10-pre10?
> 
> Tried it. Does not help.

Can you verify that 2.4.9-ac8+ shows the bug as well?

	Christoph

-- 
Whip me.  Beat me.  Make me maintain AIX.

RE: /proc/partitions hosed in 2.4.9-ac10

[Robert_Macaulay]

[-- Attachment #1: Type: text/plain, Size: 1481 bytes --]

I applied that patch to a 2.4.9ac10 kernel, and it still loops the
/proc/partitions.

-----Original Message----- 
From: Robert Macaulay 
Sent: Mon 9/17/2001 5:54 PM 
To: Erik Andersen 
Cc: Colonel; linux-kernel@vger.kernel.org 
Subject: Re: /proc/partitions hosed in 2.4.9-ac10



I have the problem as well, and my physical sector size is 512. I'll try 
the patch 

On Mon, 17 Sep 2001, Erik Andersen wrote: 

> 
> On Mon Sep 17, 2001 at 03:32:03PM -0700, Colonel wrote: 
> > 
> > Works fine here: 
> 
> But none of your devices have 2048 byte physical sectors, 
> which is the case with my MO drives, and that appears to 
> be the root of the problem, 
> 
>  -Erik 
> 
> -- 
> Erik B. Andersen   email:  andersee@debian.org, formerly of Lineo 
> --This message was written using 73% post-consumer electrons-- 
> - 
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" 
> in 
> the body of a message to majordomo@vger.kernel.org 
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
<http://vger.kernel.org/majordomo-info.html>  
> Please read the FAQ at  http://www.tux.org/lkml/
<http://www.tux.org/lkml/>  
> 
> 
> 

- 
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in 
the body of a message to majordomo@vger.kernel.org 
More majordomo info at  http://vger.kernel.org/majordomo-info.html
<http://vger.kernel.org/majordomo-info.html>  
Please read the FAQ at  http://www.tux.org/lkml/ <http://www.tux.org/lkml/>



[-- Attachment #2: Type: application/ms-tnef, Size: 5458 bytes --]

RE: /proc/partitions hosed in 2.4.9-ac10

[Robert Macaulay]

Appologies for the HTML mail. 

I applied that patch to a 2.4.9ac10 kernel, and it still loops the
/proc/partitions.


>       -----Original Message-----
>       From: Robert Macaulay
>       Sent: Mon 9/17/2001 5:54 PM
>       To: Erik Andersen
>       Cc: Colonel; linux-kernel@vger.kernel.org
>       Subject: Re: /proc/partitions hosed in 2.4.9-ac10
> 
> I have the problem as well, and my physical sector size is 512.
> I'll try
> the patch
> 
> On Mon, 17 Sep 2001, Erik Andersen wrote:
> 
> >
> > On Mon Sep 17, 2001 at 03:32:03PM -0700, Colonel wrote:
> > >
> > > Works fine here:
> >
> > But none of your devices have 2048 byte physical sectors,
> > which is the case with my MO drives, and that appears to
> > be the root of the problem,
> >
> >  -Erik
> >
> > --
> > Erik B. Andersen   email:  andersee@debian.org, formerly of
> Lineo
> > --This message was written using 73% post-consumer electrons--
> > -
> > To unsubscribe from this list: send the line "unsubscribe
> linux-kernel"
> > in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at 
> http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at  http://www.tux.org/lkml/
> >
> >
> >
> 
> -
> To unsubscribe from this list: send the line "unsubscribe
> linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at 
> http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 
> 
>