[PATCH 0/1] tools: iio: fix iio_generic_buffer

[PATCH 0/1] tools: iio: fix iio_generic_buffer

[Petre Rodan]

While executing tools/iio_generic_buffer i was getting this fault:

beagle /usr/src/linux/tools/iio # ./iio_generic_buffer -a -c 10 -N 0 -T 0
iio device number being used is 0
iio trigger number being used is 0
Auto-channels selected but some channels are already activated in sysfs
Proceeding without activating any channels
/sys/bus/iio/devices/iio:device0 mytmr
Floating point exception

The FPE was generated due to a modulo zero in size_from_channelarray()
(all channels[].bytes were zero because channels[] were never populated).

The root of the cause is the seekdir() function within
build_channel_array().

On my system I can observe a clear difference between these two functions:

    seekdir(dp, 0);
    printf("location = %lu\n", telldir(dp)); // location == 4294967295

    rewinddir(dp);
    printf("location = %lu\n", telldir(dp)); // location == 1

The attached patch fixed the bug for me.

I'm using a BeagleBone Black devboard with an up-to-date stable branch
Gentoo Linux OS on top of it.

CHOST:  armv7a-unknown-linux-gnueabihf
kernel: mainline 6.7.0-rc6
glibc:  2.37-r7:2.2::gentoo  USE="caps gd multiarch ssp static-libs -audit
    -cet -compile-locales -crypt -custom-cflags -doc -hash-sysv-compat
    -headers-only -multilib -multilib-bootstrap -nscd -perl -profile
    (-selinux) -stack-realign -suid -systemd -systemtap -test -vanilla"
    (+ contains Gentoo specific patches)

Fixes: <e58537ccce733> ("staging: iio: update example application.")
Signed-off-by: Petre Rodan <petre.rodan@subdimension.ro>

Petre Rodan (1):
  tools: iio: fix iio_generic_buffer

 tools/iio/iio_utils.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--
2.41.0

[PATCH 1/1] tools: iio: fix iio_generic_buffer

[Petre Rodan]

Fix code flow problem and floating point exception caused by improper
directory stream positioning.

Fixes: <e58537ccce733> ("staging: iio: update example application.")
Signed-off-by: Petre Rodan <petre.rodan@subdimension.ro>
---
 tools/iio/iio_utils.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/iio/iio_utils.c b/tools/iio/iio_utils.c
index 6a00a6eecaef..c5c5082cb24e 100644
--- a/tools/iio/iio_utils.c
+++ b/tools/iio/iio_utils.c
@@ -376,7 +376,7 @@ int build_channel_array(const char *device_dir, int buffer_idx,
 		goto error_close_dir;
 	}

-	seekdir(dp, 0);
+	rewinddir(dp);
 	while (ent = readdir(dp), ent) {
 		if (strcmp(ent->d_name + strlen(ent->d_name) - strlen("_en"),
 			   "_en") == 0) {
--
2.41.0

Re: [PATCH 1/1] tools: iio: fix iio_generic_buffer

[Jonathan Cameron]

On Thu,  4 Jan 2024 12:12:45 +0200
Petre Rodan <petre.rodan@subdimension.ro> wrote:

> Fix code flow problem and floating point exception caused by improper
> directory stream positioning.
> 
> Fixes: <e58537ccce733> ("staging: iio: update example application.")
> Signed-off-by: Petre Rodan <petre.rodan@subdimension.ro>
Hi Petre,

This doesn't really explain 'why' seeekdir(dp, 0) is wrong.
My assumption is that telldir() would not have returned 0 on this particular
system?  Could you confirm that and update the message to explain why
seekdir(dp, 0) is not equivalent to rewinddir(dp). 

Thanks,

Jonathan

> ---
>  tools/iio/iio_utils.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/tools/iio/iio_utils.c b/tools/iio/iio_utils.c
> index 6a00a6eecaef..c5c5082cb24e 100644
> --- a/tools/iio/iio_utils.c
> +++ b/tools/iio/iio_utils.c
> @@ -376,7 +376,7 @@ int build_channel_array(const char *device_dir, int buffer_idx,
>  		goto error_close_dir;
>  	}
> 
> -	seekdir(dp, 0);
> +	rewinddir(dp);
>  	while (ent = readdir(dp), ent) {
>  		if (strcmp(ent->d_name + strlen(ent->d_name) - strlen("_en"),
>  			   "_en") == 0) {
> --
> 2.41.0
> 

Re: [PATCH 1/1] tools: iio: fix iio_generic_buffer

[Petre Rodan]

[-- Attachment #1: Type: text/plain, Size: 1822 bytes --]


Hi Jonathan,

On Sun, Jan 07, 2024 at 03:53:33PM +0000, Jonathan Cameron wrote:
> On Thu,  4 Jan 2024 12:12:45 +0200
> Petre Rodan <petre.rodan@subdimension.ro> wrote:
> 
> > Fix code flow problem and floating point exception caused by improper
> > directory stream positioning.
> > 
> > Fixes: <e58537ccce733> ("staging: iio: update example application.")
> > Signed-off-by: Petre Rodan <petre.rodan@subdimension.ro>
> Hi Petre,
> 
> This doesn't really explain 'why' seeekdir(dp, 0) is wrong.
> My assumption is that telldir() would not have returned 0 on this particular
> system?  Could you confirm that and update the message to explain why
> seekdir(dp, 0) is not equivalent to rewinddir(dp).

it looks like there is a 32-bit specific bug in one of the glibc patches that
the stable Gentoo Linux is using [1]. telldir returns (u32)(-1) if seekdir is
used which ends up in a floating point exception.

it so happens that the rewinddir() function (which is equivalent) works
correctly. so it is a fix for Gentoo users and should not impact anyone else.

[1] https://sourceware.org/bugzilla/show_bug.cgi?id=31212

my very best regards,
peter

> Thanks,
> 
> Jonathan
> 
> > ---
> >  tools/iio/iio_utils.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/tools/iio/iio_utils.c b/tools/iio/iio_utils.c
> > index 6a00a6eecaef..c5c5082cb24e 100644
> > --- a/tools/iio/iio_utils.c
> > +++ b/tools/iio/iio_utils.c
> > @@ -376,7 +376,7 @@ int build_channel_array(const char *device_dir, int buffer_idx,
> >  		goto error_close_dir;
> >  	}
> > 
> > -	seekdir(dp, 0);
> > +	rewinddir(dp);
> >  	while (ent = readdir(dp), ent) {
> >  		if (strcmp(ent->d_name + strlen(ent->d_name) - strlen("_en"),
> >  			   "_en") == 0) {
> > --
> > 2.41.0

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH 1/1] tools: iio: fix iio_generic_buffer

[Jonathan Cameron]

On Sun, 7 Jan 2024 18:28:52 +0200
Petre Rodan <petre.rodan@subdimension.ro> wrote:

> Hi Jonathan,
> 
> On Sun, Jan 07, 2024 at 03:53:33PM +0000, Jonathan Cameron wrote:
> > On Thu,  4 Jan 2024 12:12:45 +0200
> > Petre Rodan <petre.rodan@subdimension.ro> wrote:
> >   
> > > Fix code flow problem and floating point exception caused by improper
> > > directory stream positioning.
> > > 
> > > Fixes: <e58537ccce733> ("staging: iio: update example application.")
> > > Signed-off-by: Petre Rodan <petre.rodan@subdimension.ro>  
> > Hi Petre,
> > 
> > This doesn't really explain 'why' seeekdir(dp, 0) is wrong.
> > My assumption is that telldir() would not have returned 0 on this particular
> > system?  Could you confirm that and update the message to explain why
> > seekdir(dp, 0) is not equivalent to rewinddir(dp).  
> 
> it looks like there is a 32-bit specific bug in one of the glibc patches that
> the stable Gentoo Linux is using [1]. telldir returns (u32)(-1) if seekdir is
> used which ends up in a floating point exception.
> 
> it so happens that the rewinddir() function (which is equivalent) works
> correctly. so it is a fix for Gentoo users and should not impact anyone else.
> 
> [1] https://sourceware.org/bugzilla/show_bug.cgi?id=31212
> 
> my very best regards,
> peter

I'd be happy to take the same change as a cleanup if you want to resubmit it
with a different patch description.

Thanks,

Jonathan

> 
> > Thanks,
> > 
> > Jonathan
> >   
> > > ---
> > >  tools/iio/iio_utils.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/tools/iio/iio_utils.c b/tools/iio/iio_utils.c
> > > index 6a00a6eecaef..c5c5082cb24e 100644
> > > --- a/tools/iio/iio_utils.c
> > > +++ b/tools/iio/iio_utils.c
> > > @@ -376,7 +376,7 @@ int build_channel_array(const char *device_dir, int buffer_idx,
> > >  		goto error_close_dir;
> > >  	}
> > > 
> > > -	seekdir(dp, 0);
> > > +	rewinddir(dp);
> > >  	while (ent = readdir(dp), ent) {
> > >  		if (strcmp(ent->d_name + strlen(ent->d_name) - strlen("_en"),
> > >  			   "_en") == 0) {
> > > --
> > > 2.41.0  

Re: [PATCH 1/1] tools: iio: fix iio_generic_buffer

[Petre Rodan]

[-- Attachment #1: Type: text/plain, Size: 1555 bytes --]


Hi Jonathan,

On Sun, Jan 07, 2024 at 05:00:26PM +0000, Jonathan Cameron wrote:
> > On Sun, Jan 07, 2024 at 03:53:33PM +0000, Jonathan Cameron wrote:
> > > On Thu,  4 Jan 2024 12:12:45 +0200
> > > Petre Rodan <petre.rodan@subdimension.ro> wrote:
> > >   
> > > > Fix code flow problem and floating point exception caused by improper
> > > > directory stream positioning.
> > > > 
> > > > Fixes: <e58537ccce733> ("staging: iio: update example application.")
> > > > Signed-off-by: Petre Rodan <petre.rodan@subdimension.ro>  
> > > Hi Petre,
> > > 
> > > This doesn't really explain 'why' seeekdir(dp, 0) is wrong.
> > > My assumption is that telldir() would not have returned 0 on this particular
> > > system?  Could you confirm that and update the message to explain why
> > > seekdir(dp, 0) is not equivalent to rewinddir(dp).  
> > 
> > it looks like there is a 32-bit specific bug in one of the glibc patches that
> > the stable Gentoo Linux is using [1]. telldir returns (u32)(-1) if seekdir is
> > used which ends up in a floating point exception.
> > 
> > it so happens that the rewinddir() function (which is equivalent) works
> > correctly. so it is a fix for Gentoo users and should not impact anyone else.
> > 
> > [1] https://sourceware.org/bugzilla/show_bug.cgi?id=31212
> > 
> > my very best regards,
> > peter
> 
> I'd be happy to take the same change as a cleanup if you want to resubmit it
> with a different patch description.

sure, I will resubmit it, but tomorrow morning.

best regards,
peter

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]