Skip to contentSkip to conversation

patch md-raid5-allow-for-more-than-2-31-chunks.patch added to 2.6.33-stable tree

DISCUSSION Stefan Lippers-Hollmann 3 participants 6 messages Apr 22, 2010 – Apr 23, 2010 1 message not in the archive MD RAID

lore
Share & tools
lore

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

#1 lore raw

Hi

Quoted 21 lines · On Thursday 22 April 2010, [email protected] wrote:

This is a note to let you know that we have just queued up the patch titled

Subject: md/raid5: allow for more than 2^31 chunks.

to the 2.6.33-stable tree. Its filename is

    md-raid5-allow-for-more-than-2-31-chunks.patch

A git repo of this tree can be found at
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

From 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d Mon Sep 17 00:00:00 2001
From: NeilBrown <[email protected]>
Date: Tue, 20 Apr 2010 14:13:34 +1000
Subject: md/raid5: allow for more than 2^31 chunks.

From: NeilBrown <[email protected]>

commit 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d upstream.

Quoted 7 lines

With many large drives and small chunk sizes it is possible to create a RAID5 with more than 2^31 chunks. Make sure this works.

This patch, as part of the current 2.6.33 stable queue, breaks compiling on i386 (CONFIG_LBDAF=y) for me (amd64 builds fine):

[...]
BUILD arch/x86/boot/bzImage
Root device is (254, 6)
Setup is 12700 bytes (padded to 12800 bytes). System is 2415 kB
CRC db6fa5fa
Kernel: arch/x86/boot/bzImage is ready (#1)
ERROR: "__umoddi3" [drivers/md/raid456.ko] undefined!

reverting just this patch fixes the problem for me.

Current Debian/ unstable:

ii  binutils                                            2.20.1-7                   The GNU assembler, linker and binary utilities
ii  gcc-4.4                                             4.4.3-9                    The GNU C compiler
ii  make                                                3.81-8                     An utility for Directing compilation.

Regards
Stefan Lippers-Hollmann

Quoted 63 lines
---
 drivers/md/raid5.c |   19 +++++++------------
 1 file changed, 7 insertions(+), 12 deletions(-)
drivers/md/raid5.c +7 -12
@@ -1649,8 +1649,8 @@ static sector_t raid5_compute_sector(rai
 				     int previous, int *dd_idx,
 				     struct stripe_head *sh)
 {
-	long stripe;
-	unsigned long chunk_number;
+	sector_t stripe;
+	sector_t chunk_number;
 	unsigned int chunk_offset;
 	int pd_idx, qd_idx;
 	int ddf_layout = 0;
@@ -1670,17 +1670,12 @@ static sector_t raid5_compute_sector(rai
 	 */
 	chunk_offset = sector_div(r_sector, sectors_per_chunk);
 	chunk_number = r_sector;
-	BUG_ON(r_sector != chunk_number);
 
 	/*
 	 * Compute the stripe number
 	 */
-	stripe = chunk_number / data_disks;
-
-	/*
-	 * Compute the data disk and parity disk indexes inside the stripe
-	 */
-	*dd_idx = chunk_number % data_disks;
+	stripe = chunk_number;
+	*dd_idx = sector_div(stripe, data_disks);
 
 	/*
 	 * Select the parity disk based on the user selected algorithm.
@@ -1869,14 +1864,14 @@ static sector_t compute_blocknr(struct s
 				 : conf->algorithm;
 	sector_t stripe;
 	int chunk_offset;
-	int chunk_number, dummy1, dd_idx = i;
+	sector_t chunk_number;
+	int dummy1, dd_idx = i;
 	sector_t r_sector;
 	struct stripe_head sh2;
 
 
 	chunk_offset = sector_div(new_sector, sectors_per_chunk);
 	stripe = new_sector;
-	BUG_ON(new_sector != stripe);
 
 	if (i == sh->pd_idx)
 		return 0;
@@ -1969,7 +1964,7 @@ static sector_t compute_blocknr(struct s
 	}
 
 	chunk_number = stripe * data_disks + i;
-	r_sector = (sector_t)chunk_number * sectors_per_chunk + chunk_offset;
+	r_sector = chunk_number * sectors_per_chunk + chunk_offset;
 
 	check = raid5_compute_sector(conf, r_sector,
 				     previous, &dummy1, &sh2);
#2 lore raw
Quoted 44 lines · On Thu, 22 Apr 2010 04:08:30 +0200 "Stefan Lippers-Hollmann" <[email protected]>…

Hi

Quoted 21 lines · On Thursday 22 April 2010, [email protected] wrote:

This is a note to let you know that we have just queued up the patch titled

Subject: md/raid5: allow for more than 2^31 chunks.

to the 2.6.33-stable tree. Its filename is

    md-raid5-allow-for-more-than-2-31-chunks.patch

A git repo of this tree can be found at
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

From 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d Mon Sep 17 00:00:00 2001
From: NeilBrown <[email protected]>
Date: Tue, 20 Apr 2010 14:13:34 +1000
Subject: md/raid5: allow for more than 2^31 chunks.

From: NeilBrown <[email protected]>

commit 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d upstream.

Quoted 7 lines

With many large drives and small chunk sizes it is possible to create a RAID5 with more than 2^31 chunks. Make sure this works.

This patch, as part of the current 2.6.33 stable queue, breaks compiling on i386 (CONFIG_LBDAF=y) for me (amd64 builds fine):

[...]
BUILD arch/x86/boot/bzImage
Root device is (254, 6)
Setup is 12700 bytes (padded to 12800 bytes). System is 2415 kB
CRC db6fa5fa
Kernel: arch/x86/boot/bzImage is ready (#1)
ERROR: "__umoddi3" [drivers/md/raid456.ko] undefined!

Quoted 1 line

reverting just this patch fixes the problem for me.

Thanks for testing and reporting.

If you could verify that this additional patch fixes the compile error I would really appreciate it.

Thanks,
NeilBrown

drivers/md/raid5.c +18 -17
index 20e4840..58ea0ec 100644
@@ -1650,7 +1650,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 				     int previous, int *dd_idx,
 				     struct stripe_head *sh)
 {
-	sector_t stripe;
+	sector_t stripe, stripe2;
 	sector_t chunk_number;
 	unsigned int chunk_offset;
 	int pd_idx, qd_idx;
@@ -1677,7 +1677,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 	 */
 	stripe = chunk_number;
 	*dd_idx = sector_div(stripe, data_disks);
-
+	stripe2 = stripe;
 	/*
 	 * Select the parity disk based on the user selected algorithm.
 	 */
@@ -1689,21 +1689,21 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 	case 5:
 		switch (algorithm) {
 		case ALGORITHM_LEFT_ASYMMETRIC:
-			pd_idx = data_disks - stripe % raid_disks;
+			pd_idx = data_disks - sector_div(stripe2, raid_disks);
 			if (*dd_idx >= pd_idx)
 				(*dd_idx)++;
 			break;
 		case ALGORITHM_RIGHT_ASYMMETRIC:
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			if (*dd_idx >= pd_idx)
 				(*dd_idx)++;
 			break;
 		case ALGORITHM_LEFT_SYMMETRIC:
-			pd_idx = data_disks - stripe % raid_disks;
+			pd_idx = data_disks - sector_div(stripe2, raid_disks);
 			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
 			break;
 		case ALGORITHM_RIGHT_SYMMETRIC:
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
 			break;
 		case ALGORITHM_PARITY_0:
@@ -1723,7 +1723,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 
 		switch (algorithm) {
 		case ALGORITHM_LEFT_ASYMMETRIC:
-			pd_idx = raid_disks - 1 - (stripe % raid_disks);
+			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
 			qd_idx = pd_idx + 1;
 			if (pd_idx == raid_disks-1) {
 				(*dd_idx)++;	/* Q D D D P */
@@ -1732,7 +1732,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 				(*dd_idx) += 2; /* D D P Q D */
 			break;
 		case ALGORITHM_RIGHT_ASYMMETRIC:
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			qd_idx = pd_idx + 1;
 			if (pd_idx == raid_disks-1) {
 				(*dd_idx)++;	/* Q D D D P */
@@ -1741,12 +1741,12 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 				(*dd_idx) += 2; /* D D P Q D */
 			break;
 		case ALGORITHM_LEFT_SYMMETRIC:
-			pd_idx = raid_disks - 1 - (stripe % raid_disks);
+			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
 			qd_idx = (pd_idx + 1) % raid_disks;
 			*dd_idx = (pd_idx + 2 + *dd_idx) % raid_disks;
 			break;
 		case ALGORITHM_RIGHT_SYMMETRIC:
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			qd_idx = (pd_idx + 1) % raid_disks;
 			*dd_idx = (pd_idx + 2 + *dd_idx) % raid_disks;
 			break;
@@ -1765,7 +1765,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 			/* Exactly the same as RIGHT_ASYMMETRIC, but or
 			 * of blocks for computing Q is different.
 			 */
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			qd_idx = pd_idx + 1;
 			if (pd_idx == raid_disks-1) {
 				(*dd_idx)++;	/* Q D D D P */
@@ -1780,7 +1780,8 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 			 * D D D P Q  rather than
 			 * Q D D D P
 			 */
-			pd_idx = raid_disks - 1 - ((stripe + 1) % raid_disks);
+			stripe2 += 1;
+			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
 			qd_idx = pd_idx + 1;
 			if (pd_idx == raid_disks-1) {
 				(*dd_idx)++;	/* Q D D D P */
@@ -1792,7 +1793,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 
 		case ALGORITHM_ROTATING_N_CONTINUE:
 			/* Same as left_symmetric but Q is before P */
-			pd_idx = raid_disks - 1 - (stripe % raid_disks);
+			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
 			qd_idx = (pd_idx + raid_disks - 1) % raid_disks;
 			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
 			ddf_layout = 1;
@@ -1800,27 +1801,27 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 
 		case ALGORITHM_LEFT_ASYMMETRIC_6:
 			/* RAID5 left_asymmetric, with Q on last device */
-			pd_idx = data_disks - stripe % (raid_disks-1);
+			pd_idx = data_disks - sector_div(stripe2, raid_disks-1);
 			if (*dd_idx >= pd_idx)
 				(*dd_idx)++;
 			qd_idx = raid_disks - 1;
 			break;
 
 		case ALGORITHM_RIGHT_ASYMMETRIC_6:
-			pd_idx = stripe % (raid_disks-1);
+			pd_idx = sector_div(stripe2, raid_disks-1);
 			if (*dd_idx >= pd_idx)
 				(*dd_idx)++;
 			qd_idx = raid_disks - 1;
 			break;
 
 		case ALGORITHM_LEFT_SYMMETRIC_6:
-			pd_idx = data_disks - stripe % (raid_disks-1);
+			pd_idx = data_disks - sector_div(stripe2, raid_disks-1);
 			*dd_idx = (pd_idx + 1 + *dd_idx) % (raid_disks-1);
 			qd_idx = raid_disks - 1;
 			break;
 
 		case ALGORITHM_RIGHT_SYMMETRIC_6:
-			pd_idx = stripe % (raid_disks-1);
+			pd_idx = sector_div(stripe2, raid_disks-1);
 			*dd_idx = (pd_idx + 1 + *dd_idx) % (raid_disks-1);
 			qd_idx = raid_disks - 1;
 			break;
author Tested-by
#3 lore raw

Hi

Quoted 3 lines · Neil Brown
Quoted 1 line · On Thu, 22 Apr 2010 04:08:30 +0200 "Stefan Lippers-Hollmann" <[email protected]>…

On Thursday 22 April 2010, [email protected] wrote:

[...]

Quoted 29 lines
Quoted 29 lines
Quoted 9 lines

From 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d Mon Sep 17 00:00:00 2001
From: NeilBrown <[email protected]>
Date: Tue, 20 Apr 2010 14:13:34 +1000
Subject: md/raid5: allow for more than 2^31 chunks.

From: NeilBrown <[email protected]>

commit 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d upstream.

Quoted 7 lines

With many large drives and small chunk sizes it is possible to create a RAID5 with more than 2^31 chunks. Make sure this works.

This patch, as part of the current 2.6.33 stable queue, breaks compiling on i386 (CONFIG_LBDAF=y) for me (amd64 builds fine):

[...]
BUILD arch/x86/boot/bzImage
Root device is (254, 6)
Setup is 12700 bytes (padded to 12800 bytes). System is 2415 kB
CRC db6fa5fa
Kernel: arch/x86/boot/bzImage is ready (#1)
ERROR: "__umoddi3" [drivers/md/raid456.ko] undefined!

Quoted 6 lines
Quoted 1 line

reverting just this patch fixes the problem for me.

Thanks for testing and reporting.

If you could verify that this additional patch fixes the compile error I would really appreciate it.

I can confirm that this patch on top of the original md-raid5-allow-for-more-than-2-31-chunks.patch fixes the build problem on i386 for me (amd64 continues to build fine as well).

Thank you a lot
Stefan Lippers-Hollmann

Quoted 148 lines

Thanks,
NeilBrown

drivers/md/raid5.c +18 -17
index 20e4840..58ea0ec 100644
@@ -1650,7 +1650,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 				     int previous, int *dd_idx,
 				     struct stripe_head *sh)
 {
-	sector_t stripe;
+	sector_t stripe, stripe2;
 	sector_t chunk_number;
 	unsigned int chunk_offset;
 	int pd_idx, qd_idx;
@@ -1677,7 +1677,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 	 */
 	stripe = chunk_number;
 	*dd_idx = sector_div(stripe, data_disks);
-
+	stripe2 = stripe;
 	/*
 	 * Select the parity disk based on the user selected algorithm.
 	 */
@@ -1689,21 +1689,21 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 	case 5:
 		switch (algorithm) {
 		case ALGORITHM_LEFT_ASYMMETRIC:
-			pd_idx = data_disks - stripe % raid_disks;
+			pd_idx = data_disks - sector_div(stripe2, raid_disks);
 			if (*dd_idx >= pd_idx)
 				(*dd_idx)++;
 			break;
 		case ALGORITHM_RIGHT_ASYMMETRIC:
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			if (*dd_idx >= pd_idx)
 				(*dd_idx)++;
 			break;
 		case ALGORITHM_LEFT_SYMMETRIC:
-			pd_idx = data_disks - stripe % raid_disks;
+			pd_idx = data_disks - sector_div(stripe2, raid_disks);
 			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
 			break;
 		case ALGORITHM_RIGHT_SYMMETRIC:
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
 			break;
 		case ALGORITHM_PARITY_0:
@@ -1723,7 +1723,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 
 		switch (algorithm) {
 		case ALGORITHM_LEFT_ASYMMETRIC:
-			pd_idx = raid_disks - 1 - (stripe % raid_disks);
+			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
 			qd_idx = pd_idx + 1;
 			if (pd_idx == raid_disks-1) {
 				(*dd_idx)++;	/* Q D D D P */
@@ -1732,7 +1732,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 				(*dd_idx) += 2; /* D D P Q D */
 			break;
 		case ALGORITHM_RIGHT_ASYMMETRIC:
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			qd_idx = pd_idx + 1;
 			if (pd_idx == raid_disks-1) {
 				(*dd_idx)++;	/* Q D D D P */
@@ -1741,12 +1741,12 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 				(*dd_idx) += 2; /* D D P Q D */
 			break;
 		case ALGORITHM_LEFT_SYMMETRIC:
-			pd_idx = raid_disks - 1 - (stripe % raid_disks);
+			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
 			qd_idx = (pd_idx + 1) % raid_disks;
 			*dd_idx = (pd_idx + 2 + *dd_idx) % raid_disks;
 			break;
 		case ALGORITHM_RIGHT_SYMMETRIC:
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			qd_idx = (pd_idx + 1) % raid_disks;
 			*dd_idx = (pd_idx + 2 + *dd_idx) % raid_disks;
 			break;
@@ -1765,7 +1765,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 			/* Exactly the same as RIGHT_ASYMMETRIC, but or
 			 * of blocks for computing Q is different.
 			 */
-			pd_idx = stripe % raid_disks;
+			pd_idx = sector_div(stripe2, raid_disks);
 			qd_idx = pd_idx + 1;
 			if (pd_idx == raid_disks-1) {
 				(*dd_idx)++;	/* Q D D D P */
@@ -1780,7 +1780,8 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 			 * D D D P Q  rather than
 			 * Q D D D P
 			 */
-			pd_idx = raid_disks - 1 - ((stripe + 1) % raid_disks);
+			stripe2 += 1;
+			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
 			qd_idx = pd_idx + 1;
 			if (pd_idx == raid_disks-1) {
 				(*dd_idx)++;	/* Q D D D P */
@@ -1792,7 +1793,7 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 
 		case ALGORITHM_ROTATING_N_CONTINUE:
 			/* Same as left_symmetric but Q is before P */
-			pd_idx = raid_disks - 1 - (stripe % raid_disks);
+			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
 			qd_idx = (pd_idx + raid_disks - 1) % raid_disks;
 			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
 			ddf_layout = 1;
@@ -1800,27 +1801,27 @@ static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
 
 		case ALGORITHM_LEFT_ASYMMETRIC_6:
 			/* RAID5 left_asymmetric, with Q on last device */
-			pd_idx = data_disks - stripe % (raid_disks-1);
+			pd_idx = data_disks - sector_div(stripe2, raid_disks-1);
 			if (*dd_idx >= pd_idx)
 				(*dd_idx)++;
 			qd_idx = raid_disks - 1;
 			break;
 
 		case ALGORITHM_RIGHT_ASYMMETRIC_6:
-			pd_idx = stripe % (raid_disks-1);
+			pd_idx = sector_div(stripe2, raid_disks-1);
 			if (*dd_idx >= pd_idx)
 				(*dd_idx)++;
 			qd_idx = raid_disks - 1;
 			break;
 
 		case ALGORITHM_LEFT_SYMMETRIC_6:
-			pd_idx = data_disks - stripe % (raid_disks-1);
+			pd_idx = data_disks - sector_div(stripe2, raid_disks-1);
 			*dd_idx = (pd_idx + 1 + *dd_idx) % (raid_disks-1);
 			qd_idx = raid_disks - 1;
 			break;
 
 		case ALGORITHM_RIGHT_SYMMETRIC_6:
-			pd_idx = stripe % (raid_disks-1);
+			pd_idx = sector_div(stripe2, raid_disks-1);
 			*dd_idx = (pd_idx + 1 + *dd_idx) % (raid_disks-1);
 			qd_idx = raid_disks - 1;
 			break;
#4 lore raw
Quoted 40 lines · Stefan Lippers-Hollmann

Hi

Quoted 3 lines · Neil Brown
Quoted 1 line · On Thu, 22 Apr 2010 04:08:30 +0200 "Stefan Lippers-Hollmann" <[email protected]>…

On Thursday 22 April 2010, [email protected] wrote:

[...]

Quoted 33 lines
Quoted 29 lines
Quoted 9 lines

From 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d Mon Sep 17 00:00:00 2001
From: NeilBrown <[email protected]>
Date: Tue, 20 Apr 2010 14:13:34 +1000
Subject: md/raid5: allow for more than 2^31 chunks.

From: NeilBrown <[email protected]>

commit 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d upstream.

Quoted 7 lines

With many large drives and small chunk sizes it is possible to create a RAID5 with more than 2^31 chunks. Make sure this works.

This patch, as part of the current 2.6.33 stable queue, breaks compiling on i386 (CONFIG_LBDAF=y) for me (amd64 builds fine):

[...]
BUILD arch/x86/boot/bzImage
Root device is (254, 6)
Setup is 12700 bytes (padded to 12800 bytes). System is 2415 kB
CRC db6fa5fa
Kernel: arch/x86/boot/bzImage is ready (#1)
ERROR: "__umoddi3" [drivers/md/raid456.ko] undefined!

Quoted 1 line

reverting just this patch fixes the problem for me.

Thanks for testing and reporting.

Quoted 8 lines
Quoted 2 lines

If you could verify that this additional patch fixes the compile error I would really appreciate it.

I can confirm that this patch on top of the original md-raid5-allow-for-more-than-2-31-chunks.patch fixes the build problem on i386 for me (amd64 continues to build fine as well).

Neil, care to push this to Linus?

thanks,

greg k-h

#5 lore raw
Quoted 44 lines · On Thu, 22 Apr 2010 20:11:00 -0700 Greg KH <[email protected]> wrote:
Quoted 43 lines · Stefan Lippers-Hollmann

Hi

Quoted 3 lines · Neil Brown
Quoted 1 line · On Thu, 22 Apr 2010 04:08:30 +0200 "Stefan Lippers-Hollmann" <[email protected]>…

On Thursday 22 April 2010, [email protected] wrote:

[...]

Quoted 35 lines
Quoted 29 lines
Quoted 9 lines

From 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d Mon Sep 17 00:00:00 2001
From: NeilBrown <[email protected]>
Date: Tue, 20 Apr 2010 14:13:34 +1000
Subject: md/raid5: allow for more than 2^31 chunks.

From: NeilBrown <[email protected]>

commit 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d upstream.

Quoted 7 lines

With many large drives and small chunk sizes it is possible to create a RAID5 with more than 2^31 chunks. Make sure this works.

This patch, as part of the current 2.6.33 stable queue, breaks compiling on i386 (CONFIG_LBDAF=y) for me (amd64 builds fine):

[...]
BUILD arch/x86/boot/bzImage
Root device is (254, 6)
Setup is 12700 bytes (padded to 12800 bytes). System is 2415 kB
CRC db6fa5fa
Kernel: arch/x86/boot/bzImage is ready (#1)
ERROR: "__umoddi3" [drivers/md/raid456.ko] undefined!

Quoted 1 line

reverting just this patch fixes the problem for me.

Thanks for testing and reporting.

If you could verify that this additional patch fixes the compile error I would really appreciate it.

Quoted 7 lines
Quoted 5 lines

I can confirm that this patch on top of the original md-raid5-allow-for-more-than-2-31-chunks.patch fixes the build problem on i386 for me (amd64 continues to build fine as well).

Neil, care to push this to Linus?

I've sent him a request to pull it from git://neil.brown.name/md for-linus

Thanks.
NeilBrown

author Tested-by
#6 lore raw
In reply to Greg KH

Hi

Quoted 44 lines · Greg KH
Quoted 43 lines · Stefan Lippers-Hollmann

Hi

Quoted 3 lines · Neil Brown
Quoted 1 line · On Thu, 22 Apr 2010 04:08:30 +0200 "Stefan Lippers-Hollmann" <[email protected]>…

On Thursday 22 April 2010, [email protected] wrote:

[...]

Quoted 35 lines
Quoted 29 lines
Quoted 9 lines

From 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d Mon Sep 17 00:00:00 2001
From: NeilBrown <[email protected]>
Date: Tue, 20 Apr 2010 14:13:34 +1000
Subject: md/raid5: allow for more than 2^31 chunks.

From: NeilBrown <[email protected]>

commit 35f2a591192d0a5d9f7fc696869c76f0b8e49c3d upstream.

Quoted 7 lines

With many large drives and small chunk sizes it is possible to create a RAID5 with more than 2^31 chunks. Make sure this works.

This patch, as part of the current 2.6.33 stable queue, breaks compiling on i386 (CONFIG_LBDAF=y) for me (amd64 builds fine):

[...]
BUILD arch/x86/boot/bzImage
Root device is (254, 6)
Setup is 12700 bytes (padded to 12800 bytes). System is 2415 kB
CRC db6fa5fa
Kernel: arch/x86/boot/bzImage is ready (#1)
ERROR: "__umoddi3" [drivers/md/raid456.ko] undefined!

Quoted 1 line

reverting just this patch fixes the problem for me.

Thanks for testing and reporting.

If you could verify that this additional patch fixes the compile error I would really appreciate it.

Quoted 7 lines
Quoted 5 lines

I can confirm that this patch on top of the original md-raid5-allow-for-more-than-2-31-chunks.patch fixes the build problem on i386 for me (amd64 continues to build fine as well).

Neil, care to push this to Linus?

Gitweb:     http://git.kernel.org/linus/6e3b96ed610e5a1838e62ddae9fa0c3463f235fa
Commit:     6e3b96ed610e5a1838e62ddae9fa0c3463f235fa
Parent:     35f2a591192d0a5d9f7fc696869c76f0b8e49c3d
Author:     NeilBrown <[email protected]>
AuthorDate: Fri Apr 23 07:08:28 2010 +1000
Committer:  NeilBrown <[email protected]>
CommitDate: Fri Apr 23 07:08:28 2010 +1000

md/raid5: fix previous patch.

Previous patch changes stripe and chunk_number to sector_t but mistakenly did not update all of the divisions to use sector_dev().

This patch changes all the those divisions (actually the '%' operator) to sector_div.

Regards
Stefan Lippers-Hollmann