| Message ID | 5093891D.6070302@ti.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Friday 02 November 2012 02:19 PM, Tomi Valkeinen wrote: > On 2012-11-02 08:38, Santosh Shilimkar wrote: >> Tony, >> >> On Friday 02 November 2012 04:18 AM, Tony Lindgren wrote: >>> This file has only omap_init_consistent_dma_size() >>> left that can be moved to plat-omap/dma.c. >>> >>> Signed-off-by: Tony Lindgren <tony@atomide.com> >>> --- >>> arch/arm/plat-omap/Makefile | 2 +- >>> arch/arm/plat-omap/common.c | 26 -------------------------- >>> arch/arm/plat-omap/dma.c | 8 ++++++++ >>> 3 files changed, 9 insertions(+), 27 deletions(-) >>> delete mode 100644 arch/arm/plat-omap/common.c >>> >> >> [..] >> >>> diff --git a/arch/arm/plat-omap/dma.c b/arch/arm/plat-omap/dma.c >>> index c288b76..00a3a53 100644 >>> --- a/arch/arm/plat-omap/dma.c >>> +++ b/arch/arm/plat-omap/dma.c >>> @@ -2146,6 +2146,14 @@ static struct platform_driver >>> omap_system_dma_driver = { >>> }, >>> }; >>> >>> +/* This must be called from init_early() */ >>> +void __init omap_init_consistent_dma_size(void) >>> +{ >>> +#ifdef CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE >>> + init_consistent_dma_size(CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE << 20); >>> +#endif >>> +} >>> + >> Lets not move this in DMA code since the above is really related >> to frame buffer. It reserves more DMA area for dma_alloc_coherent() >> etc than default 2 MB. Infact, we should no longer need this with >> CMA and memblock in place. >> >> Tomi, >> Can we not get rid of the above memory reservation ? > > Yes, I think so. This one is only used for the old omapfb, i.e. omap1, > and I have no means to test it out, though. But below is a patch to > remove it. I also attached the patch, as it looks like thunderbird wants > to reformat the pasted patch... I'll remove the > CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE from the omapfb driver's Kconfig file > in my tree later. > Great !! Thanks for the patches Tomi. For both the patches, feel free add. Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
* Santosh Shilimkar <santosh.shilimkar@ti.com> [121102 01:56]: > On Friday 02 November 2012 02:19 PM, Tomi Valkeinen wrote: > >On 2012-11-02 08:38, Santosh Shilimkar wrote: > >> > >>Lets not move this in DMA code since the above is really related > >>to frame buffer. It reserves more DMA area for dma_alloc_coherent() > >>etc than default 2 MB. Infact, we should no longer need this with > >>CMA and memblock in place. > >> > >>Tomi, > >>Can we not get rid of the above memory reservation ? > > > >Yes, I think so. This one is only used for the old omapfb, i.e. omap1, > >and I have no means to test it out, though. But below is a patch to > >remove it. I also attached the patch, as it looks like thunderbird wants > >to reformat the pasted patch... I'll remove the > >CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE from the omapfb driver's Kconfig file > >in my tree later. > > > Great !! > Thanks for the patches Tomi. > > For both the patches, feel free add. > Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com> Great thanks, will swap in Tomi's patch. Regards, Tony
> * Santosh Shilimkar <santosh.shilimkar@ti.com> [121102 01:56]: > > On Friday 02 November 2012 02:19 PM, Tomi Valkeinen wrote: > > >On 2012-11-02 08:38, Santosh Shilimkar wrote: > > >> > > >>Lets not move this in DMA code since the above is really related > > >>to frame buffer. It reserves more DMA area for dma_alloc_coherent() > > >>etc than default 2 MB. Infact, we should no longer need this with > > >>CMA and memblock in place. > > >> > > >>Tomi, > > >>Can we not get rid of the above memory reservation ? > > > > > >Yes, I think so. This one is only used for the old omapfb, i.e. omap1, > > >and I have no means to test it out, though. But below is a patch to > > >remove it. I also attached the patch, as it looks like thunderbird wants > > >to reformat the pasted patch... I'll remove the > > >CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE from the omapfb driver's Kconfig file > > >in my tree later. Hmm actually, is it safe to remove for omap1, or should we still keep it around for omap1? Regards, Tony
On 2012-11-02 20:54, Tony Lindgren wrote: >> * Santosh Shilimkar <santosh.shilimkar@ti.com> [121102 01:56]: >>> On Friday 02 November 2012 02:19 PM, Tomi Valkeinen wrote: >>>> On 2012-11-02 08:38, Santosh Shilimkar wrote: >>>>> >>>>> Lets not move this in DMA code since the above is really related >>>>> to frame buffer. It reserves more DMA area for dma_alloc_coherent() >>>>> etc than default 2 MB. Infact, we should no longer need this with >>>>> CMA and memblock in place. >>>>> >>>>> Tomi, >>>>> Can we not get rid of the above memory reservation ? >>>> >>>> Yes, I think so. This one is only used for the old omapfb, i.e. omap1, >>>> and I have no means to test it out, though. But below is a patch to >>>> remove it. I also attached the patch, as it looks like thunderbird wants >>>> to reformat the pasted patch... I'll remove the >>>> CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE from the omapfb driver's Kconfig file >>>> in my tree later. > > Hmm actually, is it safe to remove for omap1, or should we > still keep it around for omap1? Why wouldn't it be safe? Do you mean that CMA doesn't work on omap1, or...? I'm no expert on CMA, but as far as I can see with it's ARM generic stuff. Tomi
* Tomi Valkeinen <tomi.valkeinen@ti.com> [121105 00:06]: > On 2012-11-02 20:54, Tony Lindgren wrote: > >> * Santosh Shilimkar <santosh.shilimkar@ti.com> [121102 01:56]: > >>> On Friday 02 November 2012 02:19 PM, Tomi Valkeinen wrote: > >>>> On 2012-11-02 08:38, Santosh Shilimkar wrote: > >>>>> > >>>>> Lets not move this in DMA code since the above is really related > >>>>> to frame buffer. It reserves more DMA area for dma_alloc_coherent() > >>>>> etc than default 2 MB. Infact, we should no longer need this with > >>>>> CMA and memblock in place. > >>>>> > >>>>> Tomi, > >>>>> Can we not get rid of the above memory reservation ? > >>>> > >>>> Yes, I think so. This one is only used for the old omapfb, i.e. omap1, > >>>> and I have no means to test it out, though. But below is a patch to > >>>> remove it. I also attached the patch, as it looks like thunderbird wants > >>>> to reformat the pasted patch... I'll remove the > >>>> CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE from the omapfb driver's Kconfig file > >>>> in my tree later. > > > > Hmm actually, is it safe to remove for omap1, or should we > > still keep it around for omap1? > > Why wouldn't it be safe? Do you mean that CMA doesn't work on omap1, > or...? I'm no expert on CMA, but as far as I can see with it's ARM > generic stuff. Just wondering after your comment "This one is only used for the old omapfb, i.e. omap1". But sounds like it should no longer be needed there either if I parse that right. Regards, Tony
On 2012-11-05 17:36, Tony Lindgren wrote: > * Tomi Valkeinen <tomi.valkeinen@ti.com> [121105 00:06]: >> On 2012-11-02 20:54, Tony Lindgren wrote: >>>> * Santosh Shilimkar <santosh.shilimkar@ti.com> [121102 01:56]: >>>>> On Friday 02 November 2012 02:19 PM, Tomi Valkeinen wrote: >>>>>> On 2012-11-02 08:38, Santosh Shilimkar wrote: >>>>>>> >>>>>>> Lets not move this in DMA code since the above is really related >>>>>>> to frame buffer. It reserves more DMA area for dma_alloc_coherent() >>>>>>> etc than default 2 MB. Infact, we should no longer need this with >>>>>>> CMA and memblock in place. >>>>>>> >>>>>>> Tomi, >>>>>>> Can we not get rid of the above memory reservation ? >>>>>> >>>>>> Yes, I think so. This one is only used for the old omapfb, i.e. omap1, >>>>>> and I have no means to test it out, though. But below is a patch to >>>>>> remove it. I also attached the patch, as it looks like thunderbird wants >>>>>> to reformat the pasted patch... I'll remove the >>>>>> CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE from the omapfb driver's Kconfig file >>>>>> in my tree later. >>> >>> Hmm actually, is it safe to remove for omap1, or should we >>> still keep it around for omap1? >> >> Why wouldn't it be safe? Do you mean that CMA doesn't work on omap1, >> or...? I'm no expert on CMA, but as far as I can see with it's ARM >> generic stuff. > > Just wondering after your comment "This one is only used for the old > omapfb, i.e. omap1". But sounds like it should no longer be needed > there either if I parse that right. I meant that the code that was moved in the patch "Remove plat-omap/common.c" is only compiled and ran on omap1, when the old omapfb has been enabled in the Kconfig. It's not used on omap2+. old omapfb uses dma_alloc_*() to allocate memory, so if CMA works on omap1, I think this code in question can be removed, as done in my patch. Tomi
* Tomi Valkeinen <tomi.valkeinen@ti.com> [121105 07:50]: > On 2012-11-05 17:36, Tony Lindgren wrote: > > * Tomi Valkeinen <tomi.valkeinen@ti.com> [121105 00:06]: > >> On 2012-11-02 20:54, Tony Lindgren wrote: > >>>> * Santosh Shilimkar <santosh.shilimkar@ti.com> [121102 01:56]: > >>>>> On Friday 02 November 2012 02:19 PM, Tomi Valkeinen wrote: > >>>>>> On 2012-11-02 08:38, Santosh Shilimkar wrote: > >>>>>>> > >>>>>>> Lets not move this in DMA code since the above is really related > >>>>>>> to frame buffer. It reserves more DMA area for dma_alloc_coherent() > >>>>>>> etc than default 2 MB. Infact, we should no longer need this with > >>>>>>> CMA and memblock in place. > >>>>>>> > >>>>>>> Tomi, > >>>>>>> Can we not get rid of the above memory reservation ? > >>>>>> > >>>>>> Yes, I think so. This one is only used for the old omapfb, i.e. omap1, > >>>>>> and I have no means to test it out, though. But below is a patch to > >>>>>> remove it. I also attached the patch, as it looks like thunderbird wants > >>>>>> to reformat the pasted patch... I'll remove the > >>>>>> CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE from the omapfb driver's Kconfig file > >>>>>> in my tree later. > >>> > >>> Hmm actually, is it safe to remove for omap1, or should we > >>> still keep it around for omap1? > >> > >> Why wouldn't it be safe? Do you mean that CMA doesn't work on omap1, > >> or...? I'm no expert on CMA, but as far as I can see with it's ARM > >> generic stuff. > > > > Just wondering after your comment "This one is only used for the old > > omapfb, i.e. omap1". But sounds like it should no longer be needed > > there either if I parse that right. > > I meant that the code that was moved in the patch "Remove > plat-omap/common.c" is only compiled and ran on omap1, when the old > omapfb has been enabled in the Kconfig. It's not used on omap2+. > > old omapfb uses dma_alloc_*() to allocate memory, so if CMA works on > omap1, I think this code in question can be removed, as done in my patch. OK thanks for clarifying it. Regards, Tony
From 65c22c93928fbaaae846dd7df53343050bbcfc64 Mon Sep 17 00:00:00 2001 From: Tomi Valkeinen <tomi.valkeinen@ti.com> Date: Fri, 2 Nov 2012 10:36:13 +0200 Subject: [PATCH] ARM: OMAP: Remove omap_init_consistent_dma_size() The only thing omap_init_consistent_dma_size() does is increase the consistent DMA size if CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE is defined. Increasing the consistent DMA size should no longer be needed with CMA in place. This patch removes omap_init_consistent_dma_size() and also arch/arm/mach-omap2/io.c:omap_common_init_early() which becomes an empty function. Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com> --- arch/arm/mach-omap1/io.c | 1 - arch/arm/mach-omap2/io.c | 12 ------------ arch/arm/plat-omap/common.c | 7 ------- arch/arm/plat-omap/include/plat/dma.h | 1 - 4 files changed, 21 deletions(-) diff --git a/arch/arm/mach-omap1/io.c b/arch/arm/mach-omap1/io.c index 6a5baab..b3d0fb3 100644 --- a/arch/arm/mach-omap1/io.c +++ b/arch/arm/mach-omap1/io.c @@ -134,7 +134,6 @@ void __init omap1_init_early(void) */ omap1_clk_init(); omap1_mux_init(); - omap_init_consistent_dma_size(); } void __init omap1_init_late(void) diff --git a/arch/arm/mach-omap2/io.c b/arch/arm/mach-omap2/io.c index 4234d28..2597846 100644 --- a/arch/arm/mach-omap2/io.c +++ b/arch/arm/mach-omap2/io.c @@ -354,11 +354,6 @@ static int _set_hwmod_postsetup_state(struct omap_hwmod *oh, void *data) return omap_hwmod_set_postsetup_state(oh, *(u8 *)data); } -static void __init omap_common_init_early(void) -{ - omap_init_consistent_dma_size(); -} - static void __init omap_hwmod_init_postsetup(void) { u8 postsetup_state; @@ -379,7 +374,6 @@ void __init omap2420_init_early(void) { omap2_set_globals_242x(); omap2xxx_check_revision(); - omap_common_init_early(); omap2xxx_voltagedomains_init(); omap242x_powerdomains_init(); omap242x_clockdomains_init(); @@ -401,7 +395,6 @@ void __init omap2430_init_early(void) { omap2_set_globals_243x(); omap2xxx_check_revision(); - omap_common_init_early(); omap2xxx_voltagedomains_init(); omap243x_powerdomains_init(); omap243x_clockdomains_init(); @@ -428,7 +421,6 @@ void __init omap3_init_early(void) omap2_set_globals_3xxx(); omap3xxx_check_revision(); omap3xxx_check_features(); - omap_common_init_early(); omap3xxx_voltagedomains_init(); omap3xxx_powerdomains_init(); omap3xxx_clockdomains_init(); @@ -462,7 +454,6 @@ void __init ti81xx_init_early(void) omap2_set_globals_ti81xx(); omap3xxx_check_revision(); ti81xx_check_features(); - omap_common_init_early(); omap3xxx_voltagedomains_init(); omap3xxx_powerdomains_init(); omap3xxx_clockdomains_init(); @@ -520,7 +511,6 @@ void __init am33xx_init_early(void) omap2_set_globals_am33xx(); omap3xxx_check_revision(); ti81xx_check_features(); - omap_common_init_early(); am33xx_voltagedomains_init(); am33xx_powerdomains_init(); am33xx_clockdomains_init(); @@ -536,7 +526,6 @@ void __init omap4430_init_early(void) omap2_set_globals_443x(); omap4xxx_check_revision(); omap4xxx_check_features(); - omap_common_init_early(); omap44xx_voltagedomains_init(); omap44xx_powerdomains_init(); omap44xx_clockdomains_init(); @@ -558,7 +547,6 @@ void __init omap5_init_early(void) { omap2_set_globals_5xxx(); omap5xxx_check_revision(); - omap_common_init_early(); } #endif diff --git a/arch/arm/plat-omap/common.c b/arch/arm/plat-omap/common.c index 111315a..ab44d34 100644 --- a/arch/arm/plat-omap/common.c +++ b/arch/arm/plat-omap/common.c @@ -31,13 +31,6 @@ void __init omap_reserve(void) omap_barrier_reserve_memblock(); } -void __init omap_init_consistent_dma_size(void) -{ -#ifdef CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE - init_consistent_dma_size(CONFIG_FB_OMAP_CONSISTENT_DMA_SIZE << 20); -#endif -} - /* * Stub function for OMAP2 so that common files * continue to build when custom builds are used diff --git a/arch/arm/plat-omap/include/plat/dma.h b/arch/arm/plat-omap/include/plat/dma.h index 0a87b05..f1b2ad3 100644 --- a/arch/arm/plat-omap/include/plat/dma.h +++ b/arch/arm/plat-omap/include/plat/dma.h @@ -449,7 +449,6 @@ struct omap_system_dma_plat_info { u32 (*dma_read)(int reg, int lch); }; -extern void __init omap_init_consistent_dma_size(void); extern void omap_set_dma_priority(int lch, int dst_port, int priority); extern int omap_request_dma(int dev_id, const char *dev_name, void (*callback)(int lch, u16 ch_status, void *data), -- 1.7.10.4