The current smc91x code uses set_irq_type(). It looks like the m32r guys
worked around this by adding a !defined(__m32r__) check, but this is equally
bogus as set_irq_type() is an arm/arm26-specific function.
Trying to get this to build on sh died in the same spot, so we just back out
the m32r change and make it depend on CONFIG_ARM instead. Notably, the ppc
build would have been broken by this as well, but it doesn't seem like anyone
noticed this there yet.
Signed-off-by: Paul Mundt <[email protected]>
drivers/net/smc91x.c | 2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
===== drivers/net/smc91x.c 1.7 vs edited =====
--- 1.7/drivers/net/smc91x.c 2004-09-17 03:07:00 +03:00
+++ edited/drivers/net/smc91x.c 2004-10-12 19:05:25 +03:00
@@ -1885,7 +1885,7 @@
if (retval)
goto err_out;
-#if !defined(__m32r__)
+#ifdef CONFIG_ARM
set_irq_type(dev->irq, IRQT_RISING);
#endif
#ifdef SMC_USE_PXA_DMA
On Tue, 12 Oct 2004, Paul Mundt wrote:
> The current smc91x code uses set_irq_type(). It looks like the m32r guys
> worked around this by adding a !defined(__m32r__) check, but this is equally
> bogus as set_irq_type() is an arm/arm26-specific function.
>
> Trying to get this to build on sh died in the same spot, so we just back out
> the m32r change and make it depend on CONFIG_ARM instead. Notably, the ppc
> build would have been broken by this as well, but it doesn't seem like anyone
> noticed this there yet.
I previously suggested that architectures without set_irq_type() define
it as an empty macro, but if only ARM has that function it's probably
saner to just enable it for ARM as you did.
Nicolas
Hello, Paul,
OK. I agree with you.
From: Paul Mundt <[email protected]>
Subject: [PATCH] net: fix smc91x build for sh/ppc
Date: Tue, 12 Oct 2004 19:16:32 +0300
> The current smc91x code uses set_irq_type(). It looks like the m32r guys
> worked around this by adding a !defined(__m32r__) check, but this is equally
> bogus as set_irq_type() is an arm/arm26-specific function.
>
> Trying to get this to build on sh died in the same spot, so we just back out
> the m32r change and make it depend on CONFIG_ARM instead. Notably, the ppc
> build would have been broken by this as well, but it doesn't seem like anyone
> noticed this there yet.
>
> Signed-off-by: Paul Mundt <[email protected]>
>
> drivers/net/smc91x.c | 2 +-
> 1 files changed, 1 insertion(+), 1 deletion(-)
>
> ===== drivers/net/smc91x.c 1.7 vs edited =====
> --- 1.7/drivers/net/smc91x.c 2004-09-17 03:07:00 +03:00
> +++ edited/drivers/net/smc91x.c 2004-10-12 19:05:25 +03:00
> @@ -1885,7 +1885,7 @@
> if (retval)
> goto err_out;
>
> -#if !defined(__m32r__)
> +#ifdef CONFIG_ARM
> set_irq_type(dev->irq, IRQT_RISING);
> #endif
> #ifdef SMC_USE_PXA_DMA
-- Takata