| Message ID | 1355549964-12893-1-git-send-email-21cnbao@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Sat, 2012-12-15 at 05:39 +0000, Barry Song wrote: > From: Barry Song <Baohua.Song@csr.com> > > when using dtb to boot v2p-ca9, clcd probe will fail due to missed > the platform_data: > clcd-pl11x: probe of 10020000.clcd failed with error -22 > this patch hooks the platform_data to clcd and make lcd of v2p-ca9 > can work while booting vexpress by DTB. > Tested on qemu by "qemu-system-arm -M vexpress-a9 -m 512M -kernel > zImage -dtb ./vexpress-v2p-ca9.dtb" > > Signed-off-by: Barry Song <Baohua.Song@csr.com> Thanks for your work, however there are some problems with your code, especially if you look at the changes being merged right now for 3.8 kernel... The clocking implementation is completely changed (actually you should not need any extra CLCD information now), and as the auxdata for devices population is removed, you could use it instead of the - slightly strange to my taste - notifier-based solution. Now having said all that, I'm hoping to get comprehensive solution for DT-based VE display devices ready in time for 3.9, which will sort out the issue once for all. Pawe?
2012/12/17 Pawel Moll <pawel.moll@arm.com> > > On Sat, 2012-12-15 at 05:39 +0000, Barry Song wrote: > > From: Barry Song <Baohua.Song@csr.com> > > > > when using dtb to boot v2p-ca9, clcd probe will fail due to missed > > the platform_data: > > clcd-pl11x: probe of 10020000.clcd failed with error -22 > > this patch hooks the platform_data to clcd and make lcd of v2p-ca9 > > can work while booting vexpress by DTB. > > Tested on qemu by "qemu-system-arm -M vexpress-a9 -m 512M -kernel > > zImage -dtb ./vexpress-v2p-ca9.dtb" > > > > Signed-off-by: Barry Song <Baohua.Song@csr.com> > > Thanks for your work, however there are some problems with your code, > especially if you look at the changes being merged right now for 3.8 > kernel... The clocking implementation is completely changed (actually > you should not need any extra CLCD information now), and as the auxdata > for devices population is removed, you could use it instead of the - > slightly strange to my taste - notifier-based solution. > i see, i just tried to boot up your clcd on qemu using 3.7 release and didn't take a look at your changes for 3.8. > Now having said all that, I'm hoping to get comprehensive solution for > DT-based VE display devices ready in time for 3.9, which will sort out > the issue once for all. well, look forward to see that. but as i said, i will use 3.7 for these monthes, booting your board by dt and hope to back port yours to my 3.7 later. > > Pawe? -barry
On Mon, 2012-12-17 at 12:39 +0000, Barry Song wrote: > > Now having said all that, I'm hoping to get comprehensive solution for > > DT-based VE display devices ready in time for 3.9, which will sort out > > the issue once for all. > > well, look forward to see that. but as i said, i will use 3.7 for > these monthes, booting your board by dt and hope to back port yours to > my 3.7 later. Sure thing, if this works for you. You may also want to have a look at the Linaro VE releases, as they have working (if slightly baroque ;-) DT solution for CLCD as well. Thanks! Pawe?
On Mon, Dec 17, 2012 at 12:32:29PM +0000, Pawel Moll wrote: > On Sat, 2012-12-15 at 05:39 +0000, Barry Song wrote: > > From: Barry Song <Baohua.Song@csr.com> > > > > when using dtb to boot v2p-ca9, clcd probe will fail due to missed > > the platform_data: > > clcd-pl11x: probe of 10020000.clcd failed with error -22 > > this patch hooks the platform_data to clcd and make lcd of v2p-ca9 > > can work while booting vexpress by DTB. > > Tested on qemu by "qemu-system-arm -M vexpress-a9 -m 512M -kernel > > zImage -dtb ./vexpress-v2p-ca9.dtb" > > > > Signed-off-by: Barry Song <Baohua.Song@csr.com> > > Thanks for your work, however there are some problems with your code, > especially if you look at the changes being merged right now for 3.8 > kernel... The clocking implementation is completely changed (actually > you should not need any extra CLCD information now), and as the auxdata > for devices population is removed, you could use it instead of the - > slightly strange to my taste - notifier-based solution. > > Now having said all that, I'm hoping to get comprehensive solution for > DT-based VE display devices ready in time for 3.9, which will sort out > the issue once for all. It has? Why haven't I been copied on the CLCD driver patches then?
On Mon, 2012-12-17 at 12:55 +0000, Russell King - ARM Linux wrote: > > Now having said all that, I'm hoping to get comprehensive solution for > > DT-based VE display devices ready in time for 3.9, which will sort out > > the issue once for all. > > It has? Why haven't I been copied on the CLCD driver patches then? By "ready in time for 3.9" I meant 2-3 months timeframe. Nothing more than the Ryan Harkin's patch was sent to anyone. Pawe?
diff --git a/arch/arm/mach-vexpress/ct-ca9x4.c b/arch/arm/mach-vexpress/ct-ca9x4.c index 4f471fa..92b03e7 100644 --- a/arch/arm/mach-vexpress/ct-ca9x4.c +++ b/arch/arm/mach-vexpress/ct-ca9x4.c @@ -111,6 +111,35 @@ static struct amba_device *ct_ca9x4_amba_devs[] __initdata = { &gpio_device, }; +static int ct_ca9x4_of_amba_notify(struct notifier_block *nb, unsigned long action, + void *data) +{ + struct device *dev = data; + + /* We are only intereted in device addition */ + if (action != BUS_NOTIFY_ADD_DEVICE) + return 0; + + if (!of_find_compatible_node(NULL, NULL, "arm,vexpress,v2p-ca9")) + return 0; + + if (of_device_is_compatible(dev->of_node, "arm,pl111")) + dev->platform_data = &ct_ca9x4_clcd_data; + + return 0; +} + +static struct notifier_block ct_ca9x4_amba_nb = { + .notifier_call = ct_ca9x4_of_amba_notify, +}; + +static void ct_ca9x4_register_bus_nb(void) +{ + /* Register callbacks on OF amba device addition/removal + * to handle linking them to the right platform_data + */ + bus_register_notifier(&amba_bustype, &ct_ca9x4_amba_nb); +} static struct v2m_osc ct_osc1 = { .osc = 1, @@ -211,6 +240,7 @@ struct ct_desc ct_ca9x4_desc __initdata = { .map_io = ct_ca9x4_map_io, .init_irq = ct_ca9x4_init_irq, .init_tile = ct_ca9x4_init, + .bus_notifer = ct_ca9x4_register_bus_nb, #ifdef CONFIG_SMP .init_cpu_map = ct_ca9x4_init_cpu_map, .smp_enable = ct_ca9x4_smp_enable, diff --git a/arch/arm/mach-vexpress/include/mach/motherboard.h b/arch/arm/mach-vexpress/include/mach/motherboard.h index 1e388c7..f88982b 100644 --- a/arch/arm/mach-vexpress/include/mach/motherboard.h +++ b/arch/arm/mach-vexpress/include/mach/motherboard.h @@ -141,6 +141,7 @@ struct ct_desc { void (*init_early)(void); void (*init_irq)(void); void (*init_tile)(void); + void (*bus_notifer)(void); #ifdef CONFIG_SMP void (*init_cpu_map)(void); void (*smp_enable)(unsigned int); diff --git a/arch/arm/mach-vexpress/v2m.c b/arch/arm/mach-vexpress/v2m.c index 560e0df..e9147f8 100644 --- a/arch/arm/mach-vexpress/v2m.c +++ b/arch/arm/mach-vexpress/v2m.c @@ -431,6 +431,8 @@ static void __init v2m_clk_init(void) clk = v2m_osc_register("mb:osc1", &v2m_mb_osc1); for (i = 0; i < ARRAY_SIZE(v2m_osc1_periphs); i++) WARN_ON(clk_register_clkdev(clk, NULL, v2m_osc1_periphs[i])); + if (of_find_compatible_node(NULL, NULL, "arm,vexpress,v2p-ca9")) + WARN_ON(clk_register_clkdev(clk, NULL, "10020000.clcd")); clk = clk_register_fixed_rate(NULL, "mb:osc2", NULL, CLK_IS_ROOT, 24000000); @@ -651,7 +653,15 @@ static struct of_dev_auxdata v2m_dt_auxdata_lookup[] __initdata = { static void __init v2m_dt_init(void) { + int i; + l2x0_of_init(0x00400000, 0xfe0fffff); + + /* some devices need to hook platform_data */ + for (i = 0; i < ARRAY_SIZE(ct_descs); i++) + if (ct_descs[i]->bus_notifer) + ct_descs[i]->bus_notifer(); + of_platform_populate(NULL, of_default_bus_match_table, v2m_dt_auxdata_lookup, NULL); pm_power_off = v2m_power_off;