[Buildroot] [PATCH] bcusdk: fix endless recursion when opening USB backends

[Buildroot] [PATCH] bcusdk: fix endless recursion when opening USB backends

[Kurt Van Dijck]

The eibd usb backend implements clock_gettime, and calls pth_int_time from there.
pth_int_time will result in a call to clock_gettime, resulting in an endless
recursion.
The problem is that the USB backend shouldn't overrule clock_gettime in the first place.
Since the function does semantically and syntactically the same, the function is
place in comment alltogether.

Signed-off-by: Kurt Van Dijck <dev.kurt@vandijck-laurijssen.be>
---
 package/bcusdk/0002-linux-usbfs-clock-gettime.patch | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
 create mode 100644 package/bcusdk/0002-linux-usbfs-clock-gettime.patch

diff --git a/package/bcusdk/0002-linux-usbfs-clock-gettime.patch b/package/bcusdk/0002-linux-usbfs-clock-gettime.patch
new file mode 100644
index 0000000..6d024c6
--- /dev/null
+++ b/package/bcusdk/0002-linux-usbfs-clock-gettime.patch
@@ -0,0 +1,16 @@
+--- a/eibd/usb/linux_usbfs.c	2017-07-29 22:24:15.890087674 +0200
++++ b/eibd/usb/linux_usbfs.c	2017-07-29 22:20:11.830085100 +0200
+@@ -52,11 +52,13 @@
+ 	return 0;
+ }
+ 
++/* don't redefine clock_gettime, use the system version
+ int clock_gettime(clockid_t clk_id, struct timespec *tp)
+ {
+ 	pth_int_time (tp);
+ 	return 0;
+ }
++*/
+ 
+ /* sysfs vs usbfs:
+  * opening a usbfs node causes the device to be resumed, so we attempt to
-- 
1.8.5.rc3

[Buildroot] [PATCH] bcusdk: fix endless recursion when opening USB backends

[Thomas Petazzoni]

Hello,

On Fri, 25 Aug 2017 09:26:39 +0200, Kurt Van Dijck wrote:

> diff --git a/package/bcusdk/0002-linux-usbfs-clock-gettime.patch b/package/bcusdk/0002-linux-usbfs-clock-gettime.patch
> new file mode 100644
> index 0000000..6d024c6
> --- /dev/null
> +++ b/package/bcusdk/0002-linux-usbfs-clock-gettime.patch

All patches need to have a description + Signed-off-by. Also, since the
upstream project is using a Git repository, please use Git formatted
patches (i.e patches generated by git format-patch -N).

In addition, rather than commenting the code, just remove it entirely,
it makes the change more obvious.

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH] bcusdk: fix endless recursion when opening USB backends

[Kurt Van Dijck]

--- Original message ---
> Date: Fri, 25 Aug 2017 16:00:04 +0200
> From: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
> To: Kurt Van Dijck <dev.kurt@vandijck-laurijssen.be>
> Cc: buildroot at busybox.net
> Subject: Re: [Buildroot] [PATCH] bcusdk: fix endless recursion when opening
>  USB backends
> 
> Hello,
> 
> On Fri, 25 Aug 2017 09:26:39 +0200, Kurt Van Dijck wrote:
> 
> > diff --git a/package/bcusdk/0002-linux-usbfs-clock-gettime.patch b/package/bcusdk/0002-linux-usbfs-clock-gettime.patch
> > new file mode 100644
> > index 0000000..6d024c6
> > --- /dev/null
> > +++ b/package/bcusdk/0002-linux-usbfs-clock-gettime.patch
> 
> All patches need to have a description + Signed-off-by. Also, since the

ok. I'll redo this patch.

> upstream project is using a Git repository, please use Git formatted
> patches (i.e patches generated by git format-patch -N).

I'm not sure what you mean with 'upstream project' here? Busybox or bcusdk?
bcusdk isn't in git AFAIK.

> 
> In addition, rather than commenting the code, just remove it entirely,
> it makes the change more obvious.

ok,

Kind regards,
Kurt

[Buildroot] [PATCH] bcusdk: fix endless recursion when opening USB backends

[Thomas Petazzoni]

Hello,

On Fri, 25 Aug 2017 21:36:52 +0200, Kurt Van Dijck wrote:

> > upstream project is using a Git repository, please use Git formatted
> > patches (i.e patches generated by git format-patch -N).  
> 
> I'm not sure what you mean with 'upstream project' here? Busybox or bcusdk?

bcusdk of course.

> bcusdk isn't in git AFAIK.

https://sourceforge.net/p/bcusdk/code/ci/master/tree/ is a Git
repository.

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH] bcusdk: fix endless recursion when opening USB backends

[Kurt Van Dijck]

--- Original message ---
> Date: Fri, 25 Aug 2017 22:16:17 +0200
> From: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
> To: Kurt Van Dijck <dev.kurt@vandijck-laurijssen.be>
> Cc: buildroot at busybox.net
> Subject: Re: [Buildroot] [PATCH] bcusdk: fix endless recursion when opening
>  USB backends
> X-Mailer: Claws Mail 3.14.1 (GTK+ 2.24.31; x86_64-redhat-linux-gnu)
> 
> Hello,
> 
> On Fri, 25 Aug 2017 21:36:52 +0200, Kurt Van Dijck wrote:
> 
> > > upstream project is using a Git repository, please use Git formatted
> > > patches (i.e patches generated by git format-patch -N).  
> > 
> > I'm not sure what you mean with 'upstream project' here? Busybox or bcusdk?
> 
> bcusdk of course.
> 
> > bcusdk isn't in git AFAIK.
> 
> https://sourceforge.net/p/bcusdk/code/ci/master/tree/ is a Git
> repository.

One day, I'll switch to https://github.com/knxd/knxd anyway.

Kurt

[Buildroot] [PATCH] bcusdk: fix endless recursion when opening USB backends

[Thomas Petazzoni]

Hello,

On Fri, 25 Aug 2017 22:47:57 +0200, Kurt Van Dijck wrote:

> > https://sourceforge.net/p/bcusdk/code/ci/master/tree/ is a Git
> > repository.  
> 
> One day, I'll switch to https://github.com/knxd/knxd anyway.

Interesting. I see it's a fork of bcusdk. So perhaps we could drop the
bcusdk package at some point, and replace it with a knxd package.
Indeed, knxd seems to be actively maintained, which isn't the case of
bcusdk.

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH] bcusdk: fix endless recursion when opening USB backends

[Kurt Van Dijck]

> Hello,
> 
> On Fri, 25 Aug 2017 22:47:57 +0200, Kurt Van Dijck wrote:
> 
> > > https://sourceforge.net/p/bcusdk/code/ci/master/tree/ is a Git
> > > repository.  
> > 
> > One day, I'll switch to https://github.com/knxd/knxd anyway.
> 
> Interesting. I see it's a fork of bcusdk. So perhaps we could drop the
> bcusdk package at some point, and replace it with a knxd package.
> Indeed, knxd seems to be actively maintained, which isn't the case of
> bcusdk.

Yes. but bcusdk does it's job at the moment at my place, and knxd does not.

Kurt