[13/13] video: omap: allow building on !MMU

Message ID 1398342509-10243-14-git-send-email-peter.griffin@linaro.org
State New
Headers show

Commit Message

Peter Griffin April 24, 2014, 12:28 p.m.
From: Arnd Bergmann <arnd@arndb.de>

The omap lcdc driver has an elaborate mmap_kern function
to map the frame buffer into kernel address space as
write-combined. This uses functions that are only available
on MMU-enabled builds.

It does seem equivalent to ioremap_wc though, so we should
be able to just use that instead.

This patch is build-tested only, it needs to be run on real
hardware before applying.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Peter Griffin <peter.griffin@linaro.org>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: linux-fbdev@vger.kernel.org
Cc: linux-omap@vger.kernel.org
---
 drivers/video/fbdev/omap/lcdc.c |   28 +++++-----------------------
 1 file changed, 5 insertions(+), 23 deletions(-)

Comments

Tomi Valkeinen May 7, 2014, 9:14 a.m. | #1
On 24/04/14 15:28, Peter Griffin wrote:
> From: Arnd Bergmann <arnd@arndb.de>
> 
> The omap lcdc driver has an elaborate mmap_kern function
> to map the frame buffer into kernel address space as
> write-combined. This uses functions that are only available
> on MMU-enabled builds.
> 
> It does seem equivalent to ioremap_wc though, so we should
> be able to just use that instead.
> 
> This patch is build-tested only, it needs to be run on real
> hardware before applying.

I don't have omap1 boards, and I don't know anyone that has. I don't
even know if the omap1 fb is working or not...

Tony, any idea of omap1's fb? Are you able to test this?

 Tomi
Tony Lindgren May 8, 2014, 9:53 p.m. | #2
* Tomi Valkeinen <tomi.valkeinen@ti.com> [140507 02:15]:
> On 24/04/14 15:28, Peter Griffin wrote:
> > From: Arnd Bergmann <arnd@arndb.de>
> > 
> > The omap lcdc driver has an elaborate mmap_kern function
> > to map the frame buffer into kernel address space as
> > write-combined. This uses functions that are only available
> > on MMU-enabled builds.
> > 
> > It does seem equivalent to ioremap_wc though, so we should
> > be able to just use that instead.
> > 
> > This patch is build-tested only, it needs to be run on real
> > hardware before applying.
> 
> I don't have omap1 boards, and I don't know anyone that has. I don't
> even know if the omap1 fb is working or not...
> 
> Tony, any idea of omap1's fb? Are you able to test this?

I have three omap1 boards in my rack that I use for my boot
testing. Tried to enable framebuffer but so far no luck on
any of them. So I'm not much of a help here. Looks like the
patch should work though..

Aaro & Janusz, care to take a look at the patch in this
thread?

Regards,

Tony
--
To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Aaro Koskinen May 8, 2014, 11:04 p.m. | #3
Hi,

On Thu, May 08, 2014 at 02:53:38PM -0700, Tony Lindgren wrote:
> * Tomi Valkeinen <tomi.valkeinen@ti.com> [140507 02:15]:
> > I don't have omap1 boards, and I don't know anyone that has. I don't
> > even know if the omap1 fb is working or not...

OMAP1 fb is working very well. As a matter of fact, it's much more usable
for me than fb on OMAP2/3 - apart from N900 (on which fb is working again
with 3.15) none of my 4 other OMAP2/3 boards have a working display/fb
in mainline yet. :-(

> > Tony, any idea of omap1's fb? Are you able to test this?
> 
> I have three omap1 boards in my rack that I use for my boot
> testing. Tried to enable framebuffer but so far no luck on
> any of them. So I'm not much of a help here. Looks like the
> patch should work though..
> 
> Aaro & Janusz, care to take a look at the patch in this
> thread?

I couldn't figure out how to enter this code path. Amstrad E3 & Nokia
770 will take the alloc_fbmem() path & exit, and based on quick look
I could not see way to enter mmap_kern(). Is that dead code?

A.
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Tony Lindgren May 8, 2014, 11:17 p.m. | #4
* Aaro Koskinen <aaro.koskinen@iki.fi> [140508 16:09]:
> Hi,
> 
> On Thu, May 08, 2014 at 02:53:38PM -0700, Tony Lindgren wrote:
> > * Tomi Valkeinen <tomi.valkeinen@ti.com> [140507 02:15]:
> > > I don't have omap1 boards, and I don't know anyone that has. I don't
> > > even know if the omap1 fb is working or not...
> 
> OMAP1 fb is working very well. As a matter of fact, it's much more usable
> for me than fb on OMAP2/3 - apart from N900 (on which fb is working again
> with 3.15) none of my 4 other OMAP2/3 boards have a working display/fb
> in mainline yet. :-(

I hear you.. What do you have in the .config for omap1?
Would be nice to enable the framebuffer in omap1_defconfig?
 
> > > Tony, any idea of omap1's fb? Are you able to test this?
> > 
> > I have three omap1 boards in my rack that I use for my boot
> > testing. Tried to enable framebuffer but so far no luck on
> > any of them. So I'm not much of a help here. Looks like the
> > patch should work though..
> > 
> > Aaro & Janusz, care to take a look at the patch in this
> > thread?
> 
> I couldn't figure out how to enter this code path. Amstrad E3 & Nokia
> 770 will take the alloc_fbmem() path & exit, and based on quick look
> I could not see way to enter mmap_kern(). Is that dead code?

Could be. It may be something left from when we had code to keep
the bootloader logo displayed while booting kernel.

Regards,

Tony
--
To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Aaro Koskinen May 9, 2014, 9:21 p.m. | #5
Hi,

On Thu, May 08, 2014 at 04:17:11PM -0700, Tony Lindgren wrote:
> > OMAP1 fb is working very well. As a matter of fact, it's much more usable
> > for me than fb on OMAP2/3 - apart from N900 (on which fb is working again
> > with 3.15) none of my 4 other OMAP2/3 boards have a working display/fb
> > in mainline yet. :-(
> 
> I hear you.. What do you have in the .config for omap1?
> Would be nice to enable the framebuffer in omap1_defconfig?

Looks like the needed stuff is enabled there. Unfortunately I cannot
test the full omap1_defconfig because of size limitations.

> > > > Tony, any idea of omap1's fb? Are you able to test this?
> > > 
> > > I have three omap1 boards in my rack that I use for my boot
> > > testing. Tried to enable framebuffer but so far no luck on
> > > any of them. So I'm not much of a help here. Looks like the
> > > patch should work though..
> > > 
> > > Aaro & Janusz, care to take a look at the patch in this
> > > thread?
> > 
> > I couldn't figure out how to enter this code path. Amstrad E3 & Nokia
> > 770 will take the alloc_fbmem() path & exit, and based on quick look
> > I could not see way to enter mmap_kern(). Is that dead code?
> 
> Could be. It may be something left from when we had code to keep
> the bootloader logo displayed while booting kernel.

The support for early fbmem alloc was removed in 2011
(1e434f9318efc3dddc0c0b8d2071712668154c2b, OMAPFB: remove early mem
alloc from old omapfb), so I think we can safely delete this code.
I can prepare & test a patch for this.

A.
--
To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Tony Lindgren May 11, 2014, 3:01 p.m. | #6
* Aaro Koskinen <aaro.koskinen@iki.fi> [140509 14:26]:
> Hi,
> 
> On Thu, May 08, 2014 at 04:17:11PM -0700, Tony Lindgren wrote:
> > > OMAP1 fb is working very well. As a matter of fact, it's much more usable
> > > for me than fb on OMAP2/3 - apart from N900 (on which fb is working again
> > > with 3.15) none of my 4 other OMAP2/3 boards have a working display/fb
> > > in mainline yet. :-(
> > 
> > I hear you.. What do you have in the .config for omap1?
> > Would be nice to enable the framebuffer in omap1_defconfig?
> 
> Looks like the needed stuff is enabled there. Unfortunately I cannot
> test the full omap1_defconfig because of size limitations.

OK maybe it's just a question of loading the backlight modules.
 
> > > > > Tony, any idea of omap1's fb? Are you able to test this?
> > > > 
> > > > I have three omap1 boards in my rack that I use for my boot
> > > > testing. Tried to enable framebuffer but so far no luck on
> > > > any of them. So I'm not much of a help here. Looks like the
> > > > patch should work though..
> > > > 
> > > > Aaro & Janusz, care to take a look at the patch in this
> > > > thread?
> > > 
> > > I couldn't figure out how to enter this code path. Amstrad E3 & Nokia
> > > 770 will take the alloc_fbmem() path & exit, and based on quick look
> > > I could not see way to enter mmap_kern(). Is that dead code?
> > 
> > Could be. It may be something left from when we had code to keep
> > the bootloader logo displayed while booting kernel.
> 
> The support for early fbmem alloc was removed in 2011
> (1e434f9318efc3dddc0c0b8d2071712668154c2b, OMAPFB: remove early mem
> alloc from old omapfb), so I think we can safely delete this code.
> I can prepare & test a patch for this.

OK thanks that makes sense to me.

Regards,

Tony
--
To unsubscribe from this list: send the line "unsubscribe linux-fbdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/drivers/video/fbdev/omap/lcdc.c b/drivers/video/fbdev/omap/lcdc.c
index b52f625..40e8200 100644
--- a/drivers/video/fbdev/omap/lcdc.c
+++ b/drivers/video/fbdev/omap/lcdc.c
@@ -613,29 +613,11 @@  static void lcdc_dma_handler(u16 status, void *data)
 
 static int mmap_kern(void)
 {
-	struct vm_struct	*kvma;
-	struct vm_area_struct	vma;
-	pgprot_t		pgprot;
-	unsigned long		vaddr;
-
-	kvma = get_vm_area(lcdc.vram_size, VM_IOREMAP);
-	if (kvma == NULL) {
-		dev_err(lcdc.fbdev->dev, "can't get kernel vm area\n");
+	void __iomem *vaddr;
+
+	vaddr = ioremap_wc(lcdc.vram_phys, lcdc.vram_size);
+	if (!vaddr)
 		return -ENOMEM;
-	}
-	vma.vm_mm = &init_mm;
-
-	vaddr = (unsigned long)kvma->addr;
-	vma.vm_start = vaddr;
-	vma.vm_end = vaddr + lcdc.vram_size;
-
-	pgprot = pgprot_writecombine(pgprot_kernel);
-	if (io_remap_pfn_range(&vma, vaddr,
-			   lcdc.vram_phys >> PAGE_SHIFT,
-			   lcdc.vram_size, pgprot) < 0) {
-		dev_err(lcdc.fbdev->dev, "kernel mmap for FB memory failed\n");
-		return -EAGAIN;
-	}
 
 	lcdc.vram_virt = (void *)vaddr;
 
@@ -644,7 +626,7 @@  static int mmap_kern(void)
 
 static void unmap_kern(void)
 {
-	vunmap(lcdc.vram_virt);
+	iounmap(lcdc.vram_virt);
 }
 
 static int alloc_palette_ram(void)