[PATCH 0/3] ata: pata_ep93xx: COMPILE_TEST and ARM fixups

[PATCH 0/3] ata: pata_ep93xx: COMPILE_TEST and ARM fixups

[Rosen Penev]

This small series cleans up pata_ep93xx to allow COMPILE_TEST on
non-ARM platforms. Patch 1 avoids including ARM-specific asm headers,
patch 2 switches to unsigned long for the data variable, and patch 3
enables COMPILE_TEST.

Rosen Penev (3):
  ata: pata_ep93xx: avoid asm on non ARM
  ata: pata_ep93xx: use unsigned long for data
  ata: pata_ep93xx: add COMPILE_TEST support

 drivers/ata/Kconfig       | 2 +-
 drivers/ata/pata_ep93xx.c | 7 ++++++-
 2 files changed, 7 insertions(+), 2 deletions(-)

-- 
2.54.0

[PATCH 1/3] ata: pata_ep93xx: avoid asm on non ARM

[Rosen Penev]

The raw ARM asm delay loop prevents COMPILE_TEST builds on
non-ARM architectures. Guard it with CONFIG_ARM and provide a
cpu_relax() fallback for compilation on other architectures.

Assisted-by: Opencode:Big-pickle
Signed-off-by: Rosen Penev <rosenp@gmail.com>
---
 drivers/ata/pata_ep93xx.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/ata/pata_ep93xx.c b/drivers/ata/pata_ep93xx.c
index b2b9e0058333..f7e87d6af264 100644
--- a/drivers/ata/pata_ep93xx.c
+++ b/drivers/ata/pata_ep93xx.c
@@ -204,6 +204,7 @@ static void ep93xx_pata_enable_pio(void __iomem *base, int pio_mode)
  */
 static void ep93xx_pata_delay(unsigned long count)
 {
+#ifdef CONFIG_ARM
 	__asm__ volatile (
 		"0:\n"
 		"mov r0, r0\n"
@@ -212,6 +213,10 @@ static void ep93xx_pata_delay(unsigned long count)
 		: "=r" (count)
 		: "0" (count)
 	);
+#else
+	while (count--)
+		cpu_relax();
+#endif
 }
 
 static unsigned long ep93xx_pata_wait_for_iordy(void __iomem *base,
-- 
2.54.0

[PATCH 2/3] ata: pata_ep93xx: use unsigned long for data

[Rosen Penev]

An int is being encoded as a void pointer but that breaks on 64-bit
systems as the type needs to match pointer size.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
---
 drivers/ata/pata_ep93xx.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/ata/pata_ep93xx.c b/drivers/ata/pata_ep93xx.c
index f7e87d6af264..1663dcd00a93 100644
--- a/drivers/ata/pata_ep93xx.c
+++ b/drivers/ata/pata_ep93xx.c
@@ -977,7 +977,7 @@ static int ep93xx_pata_probe(struct platform_device *pdev)
 
 		match = soc_device_match(ep93xx_soc_table);
 		if (match)
-			ap->udma_mask = (unsigned int) match->data;
+			ap->udma_mask = (unsigned long) match->data;
 		else
 			ap->udma_mask = ATA_UDMA2;
 	}
-- 
2.54.0

[PATCH 3/3] ata: pata_ep93xx: add COMPILE_TEST support

[Rosen Penev]

Now that the build failures have been fixed, we can add COMPILE_TEST so
the buildbots can find potentially more problems.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
---
 drivers/ata/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/ata/Kconfig b/drivers/ata/Kconfig
index 462cbb01567c..6e25e9fcadb2 100644
--- a/drivers/ata/Kconfig
+++ b/drivers/ata/Kconfig
@@ -692,7 +692,7 @@ config PATA_EFAR
 
 config PATA_EP93XX
 	tristate "Cirrus Logic EP93xx PATA support"
-	depends on ARCH_EP93XX
+	depends on ARCH_EP93XX || COMPILE_TEST
 	select PATA_TIMINGS
 	help
 	  This option enables support for the PATA controller in
-- 
2.54.0

Re: [PATCH 0/3] ata: pata_ep93xx: COMPILE_TEST and ARM fixups

[Niklas Cassel]

Hello Rosen,

On Fri, May 29, 2026 at 05:36:24PM -0700, Rosen Penev wrote:
> This small series cleans up pata_ep93xx to allow COMPILE_TEST on
> non-ARM platforms. Patch 1 avoids including ARM-specific asm headers,
> patch 2 switches to unsigned long for the data variable, and patch 3
> enables COMPILE_TEST.
> 
> Rosen Penev (3):
>   ata: pata_ep93xx: avoid asm on non ARM
>   ata: pata_ep93xx: use unsigned long for data
>   ata: pata_ep93xx: add COMPILE_TEST support
> 
>  drivers/ata/Kconfig       | 2 +-
>  drivers/ata/pata_ep93xx.c | 7 ++++++-
>  2 files changed, 7 insertions(+), 2 deletions(-)

Your series looks good to me.

I do not think that you need to address the pre-existing issues reported
by Sashiko.

Intend to wait 1-2 days before picking this up, to hopefully get R-b tag
or two.


Kind regards,
Niklas

Re: [PATCH 0/3] ata: pata_ep93xx: COMPILE_TEST and ARM fixups

[Rosen Penev]

On Mon, Jun 1, 2026 at 1:54 AM Niklas Cassel <cassel@kernel.org> wrote:
>
> Hello Rosen,
>
> On Fri, May 29, 2026 at 05:36:24PM -0700, Rosen Penev wrote:
> > This small series cleans up pata_ep93xx to allow COMPILE_TEST on
> > non-ARM platforms. Patch 1 avoids including ARM-specific asm headers,
> > patch 2 switches to unsigned long for the data variable, and patch 3
> > enables COMPILE_TEST.
> >
> > Rosen Penev (3):
> >   ata: pata_ep93xx: avoid asm on non ARM
> >   ata: pata_ep93xx: use unsigned long for data
> >   ata: pata_ep93xx: add COMPILE_TEST support
> >
> >  drivers/ata/Kconfig       | 2 +-
> >  drivers/ata/pata_ep93xx.c | 7 ++++++-
> >  2 files changed, 7 insertions(+), 2 deletions(-)
>
> Your series looks good to me.
>
> I do not think that you need to address the pre-existing issues reported
> by Sashiko.
In retrospect, it might make sense to do so such that those fixes get
backported to stable kernels.
>
> Intend to wait 1-2 days before picking this up, to hopefully get R-b tag
> or two.
>
>
> Kind regards,
> Niklas

Re: [PATCH 0/3] ata: pata_ep93xx: COMPILE_TEST and ARM fixups

[Niklas Cassel]

On Mon, Jun 01, 2026 at 11:12:33AM -0700, Rosen Penev wrote:
> On Mon, Jun 1, 2026 at 1:54 AM Niklas Cassel <cassel@kernel.org> wrote:
> >
> > Hello Rosen,
> >
> > On Fri, May 29, 2026 at 05:36:24PM -0700, Rosen Penev wrote:
> > > This small series cleans up pata_ep93xx to allow COMPILE_TEST on
> > > non-ARM platforms. Patch 1 avoids including ARM-specific asm headers,
> > > patch 2 switches to unsigned long for the data variable, and patch 3
> > > enables COMPILE_TEST.
> > >
> > > Rosen Penev (3):
> > >   ata: pata_ep93xx: avoid asm on non ARM
> > >   ata: pata_ep93xx: use unsigned long for data
> > >   ata: pata_ep93xx: add COMPILE_TEST support
> > >
> > >  drivers/ata/Kconfig       | 2 +-
> > >  drivers/ata/pata_ep93xx.c | 7 ++++++-
> > >  2 files changed, 7 insertions(+), 2 deletions(-)
> >
> > Your series looks good to me.
> >
> > I do not think that you need to address the pre-existing issues reported
> > by Sashiko.
> In retrospect, it might make sense to do so such that those fixes get
> backported to stable kernels.

Oh, don't get me wrong.

If you want to fix up the pre-existing issues, we are happy to accept
patches :)

I just meant that it is not a prerequisite to get this series merged.


Kind regards,
Niklas

Re: [PATCH 0/3] ata: pata_ep93xx: COMPILE_TEST and ARM fixups

[Rosen Penev]

On Mon, Jun 1, 2026 at 11:34 AM Niklas Cassel <cassel@kernel.org> wrote:
>
> On Mon, Jun 01, 2026 at 11:12:33AM -0700, Rosen Penev wrote:
> > On Mon, Jun 1, 2026 at 1:54 AM Niklas Cassel <cassel@kernel.org> wrote:
> > >
> > > Hello Rosen,
> > >
> > > On Fri, May 29, 2026 at 05:36:24PM -0700, Rosen Penev wrote:
> > > > This small series cleans up pata_ep93xx to allow COMPILE_TEST on
> > > > non-ARM platforms. Patch 1 avoids including ARM-specific asm headers,
> > > > patch 2 switches to unsigned long for the data variable, and patch 3
> > > > enables COMPILE_TEST.
> > > >
> > > > Rosen Penev (3):
> > > >   ata: pata_ep93xx: avoid asm on non ARM
> > > >   ata: pata_ep93xx: use unsigned long for data
> > > >   ata: pata_ep93xx: add COMPILE_TEST support
> > > >
> > > >  drivers/ata/Kconfig       | 2 +-
> > > >  drivers/ata/pata_ep93xx.c | 7 ++++++-
> > > >  2 files changed, 7 insertions(+), 2 deletions(-)
> > >
> > > Your series looks good to me.
> > >
> > > I do not think that you need to address the pre-existing issues reported
> > > by Sashiko.
> > In retrospect, it might make sense to do so such that those fixes get
> > backported to stable kernels.
>
> Oh, don't get me wrong.
>
> If you want to fix up the pre-existing issues, we are happy to accept
> patches :)
Just sent. Hopefully no issues this time.
>
> I just meant that it is not a prerequisite to get this series merged.
One of those patches conflicts with the first here.
>
>
> Kind regards,
> Niklas

Re: [PATCH 0/3] ata: pata_ep93xx: COMPILE_TEST and ARM fixups

[Niklas Cassel]

On Fri, 29 May 2026 17:36:24 -0700, Rosen Penev wrote:
> This small series cleans up pata_ep93xx to allow COMPILE_TEST on
> non-ARM platforms. Patch 1 avoids including ARM-specific asm headers,
> patch 2 switches to unsigned long for the data variable, and patch 3
> enables COMPILE_TEST.
> 
> Rosen Penev (3):
>   ata: pata_ep93xx: avoid asm on non ARM
>   ata: pata_ep93xx: use unsigned long for data
>   ata: pata_ep93xx: add COMPILE_TEST support
> 
> [...]

Hello Rosen,

I applied this series:

Applied to libata/linux.git (for-7.2), thanks!

[1/3] ata: pata_ep93xx: avoid asm on non ARM
      https://git.kernel.org/libata/linux/c/cde1e86e
[2/3] ata: pata_ep93xx: use unsigned long for data
      https://git.kernel.org/libata/linux/c/8bbb394e
[3/3] ata: pata_ep93xx: add COMPILE_TEST support
      https://git.kernel.org/libata/linux/c/590da526



I can see that you have also sent out a bunch of stand-alone patches for
this driver. I'm not sure why you have an interest in this driver
specifically, but considering that a bunch of your fixes were sent out
with only a few minutes apart, I doubt that you can have tested them on
real hardware.

For these ancient drivers, we are mostly interested in real world problems
that you can reproduce on real hardware.

We don't really care that much for theoretical problems reported by AI.
If these problems were serious enough, most likely someone would have
encountered and fixed the problem by now.

One option would be to do like netdev and just remove some of these
ancient PATA drivers.


Kind regards,
Niklas