UBI Tools - problem with multiple ubi devices.

UBI Tools - problem with multiple ubi devices.

[John Smith]

In
GIT:      users/haver/mtd-utils.git
File:     ubi-utils/src/libubi.c
Function: get_ubi_info

The code counts the number of UBI volumes by looking for files called

  /sys/class/ubi0, /sys/class/ubi1, ...

but the files are actually

  /sys/class/ubi/ubi0, ...

The count starts from 1 (probably also wrong), so is usable if there is a
single UBI device. But if there is more than one UBI device, tools like
ubimkvol fail.

John

Re: UBI Tools - problem with multiple ubi devices.

[Alexander Schmidt]

On Tuesday 27 February 2007, John Smith wrote:
> In
> GIT:      users/haver/mtd-utils.git
> File:     ubi-utils/src/libubi.c
> Function: get_ubi_info
> 
> The code counts the number of UBI volumes by looking for files called
> 
>   /sys/class/ubi0, /sys/class/ubi1, ...
> 
> but the files are actually
> 
>   /sys/class/ubi/ubi0, ...
> 
> The count starts from 1 (probably also wrong), so is usable if there is a
> single UBI device. But if there is more than one UBI device, tools like
> ubimkvol fail.

Hi John,

thanks for pointing this out.

Frank, please review and commit.

Signed-off-by: Alexander Schmidt <alexs@linux.vnet.ibm.com>
---
 ubi-utils/src/libubi.c |   16 ++++++++--------
 1 files changed, 8 insertions(+), 8 deletions(-)

--- mtd-utils.orig/ubi-utils/src/libubi.c
+++ mtd-utils/ubi-utils/src/libubi.c
@@ -138,7 +138,7 @@ static int
 get_ubi_info(ubi_lib_t desc, struct ubi_info *ubi)
 {
 	int err;
-	int n = 1;
+	int dev_count = 0;
 	char *path;
 	struct stat stat;
 
@@ -147,31 +147,31 @@ get_ubi_info(ubi_lib_t desc, struct ubi_
 		return -1;
 
 	/* Calculate number of UBI devices */
-	do {
+	while (!err) {
 		char dir[20];
 
-		sprintf(&dir[0], "ubi%d", n);
-		path = mkpath(desc->sysfs_root, dir);
+		sprintf(&dir[0], "ubi%d", dev_count);
+		path = mkpath(desc->ubi_root, dir);
 		if (!path)
 			return ENOMEM;
 
 		err = lstat(path, &stat);
 		if (err == 0)
-			n += 1;
+			dev_count += 1;
 		free(path);
-	} while (err == 0);
+	}
 
 	if (errno != ENOENT)
 		return -1;
 
-	if (n == 0) {
+	if (dev_count == 0) {
 		ubi_err("no UBI devices found");
 		errno = EINVAL;
 		return -1;
 	}
 
 	errno = 0;
-	ubi->dev_count = n;
+	ubi->dev_count = dev_count;
 	return 0;
 }
 

Re: UBI Tools - problem with multiple ubi devices.

[Frank Haverkamp]

Hi Alex & John,

thanks for taking care. I included it in my git. I also reworked my git
to be based on the mtd-utils.git again. I will start there now to use
the ubi branch for the changes we want to try out before Josh can take
them into the offical mtd-utils.git/ubi.

On Wed, 2007-02-28 at 10:29 +0100, Alexander Schmidt wrote:
> On Tuesday 27 February 2007, John Smith wrote:
> > In
> > GIT:      users/haver/mtd-utils.git
> > File:     ubi-utils/src/libubi.c
> > Function: get_ubi_info
> > 
> > The code counts the number of UBI volumes by looking for files called
> > 
> >   /sys/class/ubi0, /sys/class/ubi1, ...
> > 
> > but the files are actually
> > 
> >   /sys/class/ubi/ubi0, ...
> > 
> > The count starts from 1 (probably also wrong), so is usable if there is a
> > single UBI device. But if there is more than one UBI device, tools like
> > ubimkvol fail.

John, thanks for pointing this out. Feel free to send us a patch in
addition to your bug-report. If it works we can immediately include it.
Pleaes let us know if it works for you now.

> 
> Hi John,
> 
> thanks for pointing this out.

> Frank, please review and commit.
> 
Thanks, looks good to me, included.

Frank

Re: UBI Tools - problem with multiple ubi devices.

[Josh Boyer]

On 2/28/07, Frank Haverkamp <haver@vnet.ibm.com> wrote:
> John, thanks for pointing this out. Feel free to send us a patch in
> addition to your bug-report. If it works we can immediately include it.
> Pleaes let us know if it works for you now.

John, if this patch worked for you please let us know.  I'm waiting to
pull it based on some kind of confirmation that it fixed the issue.

thx,
josh

Re: UBI Tools - problem with multiple ubi devices.

[Josh Boyer]

On 3/7/07, Josh Boyer <jwboyer@gmail.com> wrote:
> On 2/28/07, Frank Haverkamp <haver@vnet.ibm.com> wrote:
> > John, thanks for pointing this out. Feel free to send us a patch in
> > addition to your bug-report. If it works we can immediately include it.
> > Pleaes let us know if it works for you now.
>
> John, if this patch worked for you please let us know.  I'm waiting to
> pull it based on some kind of confirmation that it fixed the issue.

I committed this today.  Thanks for the patch Alexander.

josh