diff mbox series

[v2,01/19] fbdev: Prepare generic architecture helpers

Message ID 20230406143019.6709-2-tzimmermann@suse.de
State New
Headers show
Series arch: Consolidate <asm/fb.h> | expand

Commit Message

Thomas Zimmermann April 6, 2023, 2:30 p.m. UTC
Generic implementations of fb_pgprotect() and fb_is_primary_device()
have been in the source code for a long time. Prepare the header file
to make use of them.

Improve the code by using an inline function for fb_pgprotect()
and by removing include statements. The default mode set by
fb_pgprotect() is now writecombine, which is what most platforms
want.

Symbols are protected by preprocessor guards. Architectures that
provide a symbol need to define a preprocessor token of the same
name and value. Otherwise the header file will provide a generic
implementation. This pattern has been taken from <asm/io.h>.

v2:
	*  use writecombine mappings by default (Arnd)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
---
 include/asm-generic/fb.h | 23 +++++++++++++++++++++--
 1 file changed, 21 insertions(+), 2 deletions(-)

Comments

Geert Uytterhoeven April 11, 2023, 8:08 a.m. UTC | #1
Hi Thomas,

On Thu, Apr 6, 2023 at 4:30 PM Thomas Zimmermann <tzimmermann@suse.de> wrote:
> Generic implementations of fb_pgprotect() and fb_is_primary_device()
> have been in the source code for a long time. Prepare the header file
> to make use of them.
>
> Improve the code by using an inline function for fb_pgprotect()
> and by removing include statements. The default mode set by
> fb_pgprotect() is now writecombine, which is what most platforms
> want.
>
> Symbols are protected by preprocessor guards. Architectures that
> provide a symbol need to define a preprocessor token of the same
> name and value. Otherwise the header file will provide a generic
> implementation. This pattern has been taken from <asm/io.h>.
>
> v2:
>         *  use writecombine mappings by default (Arnd)
>
> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>

Thanks for your patch!

> --- a/include/asm-generic/fb.h
> +++ b/include/asm-generic/fb.h
> @@ -1,13 +1,32 @@
>  /* SPDX-License-Identifier: GPL-2.0 */
> +
>  #ifndef __ASM_GENERIC_FB_H_
>  #define __ASM_GENERIC_FB_H_
> -#include <linux/fb.h>
>
> -#define fb_pgprotect(...) do {} while (0)
> +/*
> + * Only include this header file from your architecture's <asm/fb.h>.
> + */
> +
> +#include <asm/page.h>
> +
> +struct fb_info;
> +struct file;
> +
> +#ifndef fb_pgprotect
> +#define fb_pgprotect fb_pgprotect
> +static inline void fb_pgprotect(struct file *file, struct vm_area_struct *vma,
> +                               unsigned long off)

Does this affect any noMMU platforms that relied on fb_pgprotect()
doing nothing before?
Perhaps the body below should be protected by "#ifdef CONFIG_MMU"?

> +{
> +       vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);

Shouldn't this use the pgprot_val() wrapper?

> +}
> +#endif

Gr{oetje,eeting}s,

                        Geert
Thomas Zimmermann April 17, 2023, 9:03 a.m. UTC | #2
Hi Geert

Am 11.04.23 um 10:08 schrieb Geert Uytterhoeven:
> Hi Thomas,
> 
> On Thu, Apr 6, 2023 at 4:30 PM Thomas Zimmermann <tzimmermann@suse.de> wrote:
>> Generic implementations of fb_pgprotect() and fb_is_primary_device()
>> have been in the source code for a long time. Prepare the header file
>> to make use of them.
>>
>> Improve the code by using an inline function for fb_pgprotect()
>> and by removing include statements. The default mode set by
>> fb_pgprotect() is now writecombine, which is what most platforms
>> want.
>>
>> Symbols are protected by preprocessor guards. Architectures that
>> provide a symbol need to define a preprocessor token of the same
>> name and value. Otherwise the header file will provide a generic
>> implementation. This pattern has been taken from <asm/io.h>.
>>
>> v2:
>>          *  use writecombine mappings by default (Arnd)
>>
>> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
> 
> Thanks for your patch!
> 
>> --- a/include/asm-generic/fb.h
>> +++ b/include/asm-generic/fb.h
>> @@ -1,13 +1,32 @@
>>   /* SPDX-License-Identifier: GPL-2.0 */
>> +
>>   #ifndef __ASM_GENERIC_FB_H_
>>   #define __ASM_GENERIC_FB_H_
>> -#include <linux/fb.h>
>>
>> -#define fb_pgprotect(...) do {} while (0)
>> +/*
>> + * Only include this header file from your architecture's <asm/fb.h>.
>> + */
>> +
>> +#include <asm/page.h>
>> +
>> +struct fb_info;
>> +struct file;
>> +
>> +#ifndef fb_pgprotect
>> +#define fb_pgprotect fb_pgprotect
>> +static inline void fb_pgprotect(struct file *file, struct vm_area_struct *vma,
>> +                               unsigned long off)
> 
> Does this affect any noMMU platforms that relied on fb_pgprotect()
> doing nothing before?
> Perhaps the body below should be protected by "#ifdef CONFIG_MMU"?

I cannot conclusively answer this question, but I did some grep'ing 
('git grep ndef | grep CONFIG_MMU'):

Only the architectures in this patchset provide <asm/fb.h> but nothing 
anywhere uses <asm-generic/fb.h> yet. And of those architectures, only 
arm and m68k have !CONFIG_MMU cases. Those are handled in the rsp 
patches. I think we're good.

> 
>> +{
>> +       vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
> 
> Shouldn't this use the pgprot_val() wrapper?

No, I think. Grep'ing for vm_page_prot, I'm not seeing it being used in 
such assignments.

Best regards
Thomas

> 
>> +}
>> +#endif
> 
> Gr{oetje,eeting}s,
> 
>                          Geert
>
Arnd Bergmann April 17, 2023, 9:22 a.m. UTC | #3
On Mon, Apr 17, 2023, at 11:03, Thomas Zimmermann wrote:
> Am 11.04.23 um 10:08 schrieb Geert Uytterhoeven:
>> On Thu, Apr 6, 2023 at 4:30 PM Thomas Zimmermann <tzimmermann@suse.de> wrote:

>>> +#ifndef fb_pgprotect
>>> +#define fb_pgprotect fb_pgprotect
>>> +static inline void fb_pgprotect(struct file *file, struct vm_area_struct *vma,
>>> +                               unsigned long off)
>> 
>> Does this affect any noMMU platforms that relied on fb_pgprotect()
>> doing nothing before?
>> Perhaps the body below should be protected by "#ifdef CONFIG_MMU"?
>
> I cannot conclusively answer this question, but I did some grep'ing 
> ('git grep ndef | grep CONFIG_MMU'):
>
> Only the architectures in this patchset provide <asm/fb.h> but nothing 
> anywhere uses <asm-generic/fb.h> yet. And of those architectures, only 
> arm and m68k have !CONFIG_MMU cases. Those are handled in the rsp 
> patches. I think we're good.

Agreed. The generic version is just a more elaborate way to do
nothing here, as the 

     vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);

line on nommu just turns into a self-assignment of the same member
that was set the line before.

     Arnd
diff mbox series

Patch

diff --git a/include/asm-generic/fb.h b/include/asm-generic/fb.h
index f9f18101ed36..631d97c507ca 100644
--- a/include/asm-generic/fb.h
+++ b/include/asm-generic/fb.h
@@ -1,13 +1,32 @@ 
 /* SPDX-License-Identifier: GPL-2.0 */
+
 #ifndef __ASM_GENERIC_FB_H_
 #define __ASM_GENERIC_FB_H_
-#include <linux/fb.h>
 
-#define fb_pgprotect(...) do {} while (0)
+/*
+ * Only include this header file from your architecture's <asm/fb.h>.
+ */
+
+#include <asm/page.h>
+
+struct fb_info;
+struct file;
+
+#ifndef fb_pgprotect
+#define fb_pgprotect fb_pgprotect
+static inline void fb_pgprotect(struct file *file, struct vm_area_struct *vma,
+				unsigned long off)
+{
+	vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
+}
+#endif
 
+#ifndef fb_is_primary_device
+#define fb_is_primary_device fb_is_primary_device
 static inline int fb_is_primary_device(struct fb_info *info)
 {
 	return 0;
 }
+#endif
 
 #endif /* __ASM_GENERIC_FB_H_ */