| Message ID | CANq1E4RAG+sJWbMShs5te89NAEQ6d=JRRry5NxG690VJTc4zmQ@mail.gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Sat, Sep 7, 2013 at 2:25 PM, David Herrmann <dh.herrmann@gmail.com> wrote: > On Sat, Sep 7, 2013 at 1:47 PM, Tom Gundersen <teg@jklm.no> wrote: >> A related question: is it expected that simplefb should be >> significantly slower than efifb, or is that something worth looking >> into? My boot with simplefb is roughly five seconds slower than with >> efifb. Coincidentally, I notice the same (or similar slowdown) with >> inteldrmfb when I see the oops (but not otherwise). > > That is probably related to the missing write-combine tag in ioremap. > Stephen, any objections to this attached patch? > Tom, if this solves the speed-issues, I will send it out once I get home. This solves the speed issues. Thanks! Feel free to add my tested-by. Cheers, Tom > From dbfb8e12166d494cd60823cbe84134d5d1a73ec8 Mon Sep 17 00:00:00 2001 > From: David Herrmann <dh.herrmann@gmail.com> > Date: Sat, 7 Sep 2013 14:22:01 +0200 > Subject: [PATCH] devm/simplefb: introduce and use devm_ioremap_wc() > > We want to use devm_ioremap_nocache() or even devm_ioremap_wc() to speed > up fbdev writes _a lot_. As devm_ioremap_wc() doesn't exist, yet, > introduce it along the way. Note that ioremap_wc() is aliases to > ioremap_nocache() in asm-generic/{io,iomem}.h so we can safely expect all > architectures to either provide it or use the same alias. > > Reported-by: Tom Gundersen <teg@jklm.no> > Signed-off-by: David Herrmann <dh.herrmann@gmail.com> > Cc: Stephen Warren <swarren@wwwdotorg.org> > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > Cc: Andrew Morton <akpm@linux-foundation.org> > --- > drivers/video/simplefb.c | 4 ++-- > include/linux/io.h | 2 ++ > lib/devres.c | 28 ++++++++++++++++++++++++++++ > 3 files changed, 32 insertions(+), 2 deletions(-) > > diff --git a/drivers/video/simplefb.c b/drivers/video/simplefb.c > index 8d78106..a29f1c4 100644 > --- a/drivers/video/simplefb.c > +++ b/drivers/video/simplefb.c > @@ -212,8 +212,8 @@ static int simplefb_probe(struct platform_device *pdev) > > info->fbops = &simplefb_ops; > info->flags = FBINFO_DEFAULT | FBINFO_MISC_FIRMWARE; > - info->screen_base = devm_ioremap(&pdev->dev, info->fix.smem_start, > - info->fix.smem_len); > + info->screen_base = devm_ioremap_wc(&pdev->dev, info->fix.smem_start, > + info->fix.smem_len); > if (!info->screen_base) { > framebuffer_release(info); > return -ENODEV; > diff --git a/include/linux/io.h b/include/linux/io.h > index f4f42fa..c529410 100644 > --- a/include/linux/io.h > +++ b/include/linux/io.h > @@ -62,6 +62,8 @@ void __iomem *devm_ioremap(struct device *dev, > resource_size_t offset, > unsigned long size); > void __iomem *devm_ioremap_nocache(struct device *dev, resource_size_t offset, > unsigned long size); > +void __iomem *devm_ioremap_wc(struct device *dev, resource_size_t offset, > + unsigned long size); > void devm_iounmap(struct device *dev, void __iomem *addr); > int check_signature(const volatile void __iomem *io_addr, > const unsigned char *signature, int length); > diff --git a/lib/devres.c b/lib/devres.c > index 8235331..34af7a9 100644 > --- a/lib/devres.c > +++ b/lib/devres.c > @@ -72,6 +72,34 @@ void __iomem *devm_ioremap_nocache(struct device > *dev, resource_size_t offset, > EXPORT_SYMBOL(devm_ioremap_nocache); > > /** > + * devm_ioremap_wc - Managed ioremap_wc() > + * @dev: Generic device to remap IO address for > + * @offset: BUS offset to map > + * @size: Size of map > + * > + * Managed ioremap_wc(). Map is automatically unmapped on driver detach. > + */ > +void __iomem *devm_ioremap_wc(struct device *dev, resource_size_t offset, > + unsigned long size) > +{ > + void __iomem **ptr, *addr; > + > + ptr = devres_alloc(devm_ioremap_release, sizeof(*ptr), GFP_KERNEL); > + if (!ptr) > + return NULL; > + > + addr = ioremap_wc(offset, size); > + if (addr) { > + *ptr = addr; > + devres_add(dev, ptr); > + } else > + devres_free(ptr); > + > + return addr; > +} > +EXPORT_SYMBOL(devm_ioremap_wc); > + > +/** > * devm_iounmap - Managed iounmap() > * @dev: Generic device to unmap for > * @addr: Address to unmap > -- > 1.8.4 -- 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
On Sat, Sep 7, 2013 at 9:25 PM, David Herrmann <dh.herrmann@gmail.com> wrote: > From dbfb8e12166d494cd60823cbe84134d5d1a73ec8 Mon Sep 17 00:00:00 2001 > From: David Herrmann <dh.herrmann@gmail.com> > Date: Sat, 7 Sep 2013 14:22:01 +0200 > Subject: [PATCH] devm/simplefb: introduce and use devm_ioremap_wc() > > We want to use devm_ioremap_nocache() or even devm_ioremap_wc() to speed > up fbdev writes _a lot_. As devm_ioremap_wc() doesn't exist, yet, > introduce it along the way. Note that ioremap_wc() is aliases to > ioremap_nocache() in asm-generic/{io,iomem}.h so we can safely expect all > architectures to either provide it or use the same alias. > > Reported-by: Tom Gundersen <teg@jklm.no> > Signed-off-by: David Herrmann <dh.herrmann@gmail.com> > Cc: Stephen Warren <swarren@wwwdotorg.org> > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > Cc: Andrew Morton <akpm@linux-foundation.org> Tried this on SHIELD, and it dramatically increases performance. Tested-by: Alexandre Courbot <acourbot@nvidia.com> Alex. -- 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
On Mon, Sep 9, 2013 at 11:36 AM, Alexandre Courbot <gnurou@gmail.com> wrote: > On Sat, Sep 7, 2013 at 9:25 PM, David Herrmann <dh.herrmann@gmail.com> wrote: >> From dbfb8e12166d494cd60823cbe84134d5d1a73ec8 Mon Sep 17 00:00:00 2001 >> From: David Herrmann <dh.herrmann@gmail.com> >> Date: Sat, 7 Sep 2013 14:22:01 +0200 >> Subject: [PATCH] devm/simplefb: introduce and use devm_ioremap_wc() >> >> We want to use devm_ioremap_nocache() or even devm_ioremap_wc() to speed >> up fbdev writes _a lot_. As devm_ioremap_wc() doesn't exist, yet, >> introduce it along the way. Note that ioremap_wc() is aliases to >> ioremap_nocache() in asm-generic/{io,iomem}.h so we can safely expect all >> architectures to either provide it or use the same alias. >> >> Reported-by: Tom Gundersen <teg@jklm.no> >> Signed-off-by: David Herrmann <dh.herrmann@gmail.com> >> Cc: Stephen Warren <swarren@wwwdotorg.org> >> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> >> Cc: Andrew Morton <akpm@linux-foundation.org> > > Tried this on SHIELD, and it dramatically increases performance. > > Tested-by: Alexandre Courbot <acourbot@nvidia.com> Maybe the patch should be split into two though, one that adds devm_ioremap_wc() and another that takes advantage of it for simplefb. Alex. -- 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
Hi On Mon, Sep 9, 2013 at 4:40 AM, Alexandre Courbot <gnurou@gmail.com> wrote: > On Mon, Sep 9, 2013 at 11:36 AM, Alexandre Courbot <gnurou@gmail.com> wrote: >> On Sat, Sep 7, 2013 at 9:25 PM, David Herrmann <dh.herrmann@gmail.com> wrote: >>> From dbfb8e12166d494cd60823cbe84134d5d1a73ec8 Mon Sep 17 00:00:00 2001 >>> From: David Herrmann <dh.herrmann@gmail.com> >>> Date: Sat, 7 Sep 2013 14:22:01 +0200 >>> Subject: [PATCH] devm/simplefb: introduce and use devm_ioremap_wc() >>> >>> We want to use devm_ioremap_nocache() or even devm_ioremap_wc() to speed >>> up fbdev writes _a lot_. As devm_ioremap_wc() doesn't exist, yet, >>> introduce it along the way. Note that ioremap_wc() is aliases to >>> ioremap_nocache() in asm-generic/{io,iomem}.h so we can safely expect all >>> architectures to either provide it or use the same alias. >>> >>> Reported-by: Tom Gundersen <teg@jklm.no> >>> Signed-off-by: David Herrmann <dh.herrmann@gmail.com> >>> Cc: Stephen Warren <swarren@wwwdotorg.org> >>> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> >>> Cc: Andrew Morton <akpm@linux-foundation.org> >> >> Tried this on SHIELD, and it dramatically increases performance. >> >> Tested-by: Alexandre Courbot <acourbot@nvidia.com> > > Maybe the patch should be split into two though, one that adds > devm_ioremap_wc() and another that takes advantage of it for simplefb. Thanks for testing! I added your tested-by. I also changed the patches slightly to no longer use devm_*. The *_wc patch is now split off, too. I will send it later today. Thanks David -- 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
On 09/07/2013 06:25 AM, David Herrmann wrote: > On Sat, Sep 7, 2013 at 1:47 PM, Tom Gundersen <teg@jklm.no> wrote: ... >> A related question: is it expected that simplefb should be >> significantly slower than efifb, or is that something worth looking >> into? My boot with simplefb is roughly five seconds slower than with >> efifb. Coincidentally, I notice the same (or similar slowdown) with >> inteldrmfb when I see the oops (but not otherwise). > > That is probably related to the missing write-combine tag in ioremap. > Stephen, any objections to this attached patch? > Tom, if this solves the speed-issues, I will send it out once I get home. > > Thanks > David > > (Patch also attached in case of new-lines issues) > > From dbfb8e12166d494cd60823cbe84134d5d1a73ec8 Mon Sep 17 00:00:00 2001 > From: David Herrmann <dh.herrmann@gmail.com> > Date: Sat, 7 Sep 2013 14:22:01 +0200 > Subject: [PATCH] devm/simplefb: introduce and use devm_ioremap_wc() > > We want to use devm_ioremap_nocache() or even devm_ioremap_wc() to speed > up fbdev writes _a lot_. As devm_ioremap_wc() doesn't exist, yet, > introduce it along the way. Note that ioremap_wc() is aliases to > ioremap_nocache() in asm-generic/{io,iomem}.h so we can safely expect all > architectures to either provide it or use the same alias. I'm fine with this so long as wc mappings are always possible, or automatically fall back to uc if not. This certainly works on Tegra, and I'll try to remember to test it on Raspberry Pi tonight. -- 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
On 09/07/2013 06:25 AM, David Herrmann wrote: > On Sat, Sep 7, 2013 at 1:47 PM, Tom Gundersen <teg@jklm.no> wrote: ... >> A related question: is it expected that simplefb should be >> significantly slower than efifb, or is that something worth looking >> into? My boot with simplefb is roughly five seconds slower than with >> efifb. Coincidentally, I notice the same (or similar slowdown) with >> inteldrmfb when I see the oops (but not otherwise). > > That is probably related to the missing write-combine tag in ioremap. > Stephen, any objections to this attached patch? > Tom, if this solves the speed-issues, I will send it out once I get home. > > Thanks > David > > (Patch also attached in case of new-lines issues) > > From dbfb8e12166d494cd60823cbe84134d5d1a73ec8 Mon Sep 17 00:00:00 2001 > From: David Herrmann <dh.herrmann@gmail.com> > Date: Sat, 7 Sep 2013 14:22:01 +0200 > Subject: [PATCH] devm/simplefb: introduce and use devm_ioremap_wc() > > We want to use devm_ioremap_nocache() or even devm_ioremap_wc() to speed > up fbdev writes _a lot_. As devm_ioremap_wc() doesn't exist, yet, > introduce it along the way. Note that ioremap_wc() is aliases to > ioremap_nocache() in asm-generic/{io,iomem}.h so we can safely expect all > architectures to either provide it or use the same alias. OK, that works fine on the Raspberry Pi too, so, Tested-by: Stephen Warren <swarren@wwwdotorg.org> -- 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
Hi On Tue, Sep 10, 2013 at 4:34 AM, Stephen Warren <swarren@wwwdotorg.org> wrote: > On 09/07/2013 06:25 AM, David Herrmann wrote: >> On Sat, Sep 7, 2013 at 1:47 PM, Tom Gundersen <teg@jklm.no> wrote: > ... >>> A related question: is it expected that simplefb should be >>> significantly slower than efifb, or is that something worth looking >>> into? My boot with simplefb is roughly five seconds slower than with >>> efifb. Coincidentally, I notice the same (or similar slowdown) with >>> inteldrmfb when I see the oops (but not otherwise). >> >> That is probably related to the missing write-combine tag in ioremap. >> Stephen, any objections to this attached patch? >> Tom, if this solves the speed-issues, I will send it out once I get home. >> >> Thanks >> David >> >> (Patch also attached in case of new-lines issues) >> >> From dbfb8e12166d494cd60823cbe84134d5d1a73ec8 Mon Sep 17 00:00:00 2001 >> From: David Herrmann <dh.herrmann@gmail.com> >> Date: Sat, 7 Sep 2013 14:22:01 +0200 >> Subject: [PATCH] devm/simplefb: introduce and use devm_ioremap_wc() >> >> We want to use devm_ioremap_nocache() or even devm_ioremap_wc() to speed >> up fbdev writes _a lot_. As devm_ioremap_wc() doesn't exist, yet, >> introduce it along the way. Note that ioremap_wc() is aliases to >> ioremap_nocache() in asm-generic/{io,iomem}.h so we can safely expect all >> architectures to either provide it or use the same alias. > > OK, that works fine on the Raspberry Pi too, so, > Tested-by: Stephen Warren <swarren@wwwdotorg.org> Back from the US and all patches sent out. Sorry for the delay. If I missed something, please let me know. Thanks for your feedback! David -- 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
diff --git a/drivers/video/simplefb.c b/drivers/video/simplefb.c index 8d78106..a29f1c4 100644 --- a/drivers/video/simplefb.c +++ b/drivers/video/simplefb.c @@ -212,8 +212,8 @@ static int simplefb_probe(struct platform_device *pdev) info->fbops = &simplefb_ops; info->flags = FBINFO_DEFAULT | FBINFO_MISC_FIRMWARE; - info->screen_base = devm_ioremap(&pdev->dev, info->fix.smem_start, - info->fix.smem_len); + info->screen_base = devm_ioremap_wc(&pdev->dev, info->fix.smem_start, + info->fix.smem_len); if (!info->screen_base) { framebuffer_release(info); return -ENODEV; diff --git a/include/linux/io.h b/include/linux/io.h index f4f42fa..c529410 100644 --- a/include/linux/io.h +++ b/include/linux/io.h @@ -62,6 +62,8 @@ void __iomem *devm_ioremap(struct device *dev, resource_size_t offset, unsigned long size); void __iomem *devm_ioremap_nocache(struct device *dev, resource_size_t offset, unsigned long size); +void __iomem *devm_ioremap_wc(struct device *dev, resource_size_t offset, + unsigned long size); void devm_iounmap(struct device *dev, void __iomem *addr); int check_signature(const volatile void __iomem *io_addr, const unsigned char *signature, int length); diff --git a/lib/devres.c b/lib/devres.c index 8235331..34af7a9 100644 --- a/lib/devres.c +++ b/lib/devres.c @@ -72,6 +72,34 @@ void __iomem *devm_ioremap_nocache(struct device *dev, resource_size_t offset, EXPORT_SYMBOL(devm_ioremap_nocache); /** + * devm_ioremap_wc - Managed ioremap_wc() + * @dev: Generic device to remap IO address for + * @offset: BUS offset to map + * @size: Size of map + * + * Managed ioremap_wc(). Map is automatically unmapped on driver detach. + */ +void __iomem *devm_ioremap_wc(struct device *dev, resource_size_t offset, + unsigned long size) +{ + void __iomem **ptr, *addr; + + ptr = devres_alloc(devm_ioremap_release, sizeof(*ptr), GFP_KERNEL); + if (!ptr) + return NULL; + + addr = ioremap_wc(offset, size); + if (addr) { + *ptr = addr; + devres_add(dev, ptr); + } else + devres_free(ptr); + + return addr; +} +EXPORT_SYMBOL(devm_ioremap_wc); + +/** * devm_iounmap - Managed iounmap() * @dev: Generic device to unmap for