fixes for broken PM init with multi-OMAP
Subsystems (3)
cover fixes for broken PM init with multi-OMAP
multi-OMAP builds are broken in various ways due to conditional code. Here is a small series of fixes.
Build tested: omap3_defconfig, omap_4430sdp_defconfig Boot tested: omap3_defconfig on OMAP3EVM
Boot tested: omap3_defconfig on OMAP3430SDP (Thanks, Santosh) Boot tested: omap3_defconfig on OMAP4430SDP (Thanks, Santosh)
Applies on current l-o master branch.
Updates from RFC patch:
- removed 'FIXME' around conditional SDRC execution for OMAP4 since OMAP4 has no SDRC.
Kevin Hilman (2):
OMAP2: cpu_is_omap2*: fix compile-time removal of unused code.
OMAP4: fix temporary hacks that break multi-omap PM
arch/arm/mach-omap2/io.c | 13 ++++++------- arch/arm/plat-omap/include/plat/cpu.h | 8 ++++++-- 2 files changed, 12 insertions(+), 9 deletions(-)
PATCH 1/2 OMAP2: cpu_is_omap2*: fix compile-time removal of unused code.
Currently if omap2420 is defined but not omap2430, cpu_is_omap2430() is still defined as a macro, instead of #define'd to zero. This results in conditional cpu_is_omap2430() code still being compiled, and leads to possible compile/link errors. In particular for hwmod init.
To fix, add extra #ifdefs to CPU check macros to ensure that the is_omap* macros are zero for each OMAP2 if they are not configured into the kernel.
--- arch/arm/plat-omap/include/plat/cpu.h | 8 ++++++-- 1 files changed, 6 insertions(+), 2 deletions(-)
arch/arm/plat-omap/include/plat/cpu.h +6 -2
@@ -167,10 +167,14 @@ IS_OMAP_SUBCLASS(443x, 0x443)#if defined(MULTI_OMAP2 )# if defined(CONFIG_ARCH_OMAP2 )# undef cpu_is_omap24xx -# undef cpu_is_omap242x -# undef cpu_is_omap243x# define cpu_is_omap24xx () is_omap24xx() +# endif +# if defined (CONFIG_ARCH_OMAP2420 ) +# undef cpu_is_omap242x# define cpu_is_omap242x () is_omap242x() +# endif +# if defined (CONFIG_ARCH_OMAP2430 ) +# undef cpu_is_omap243x# define cpu_is_omap243x () is_omap243x()# endif # if defined(CONFIG_ARCH_OMAP3 )
Signature · 1 line
1.7.0.2
Tony Lindgren
maintainerPATCH 1/2 OMAP2: cpu_is_omap2*: fix compile-time removal of
This patch has been applied to the linux-omap
by youw fwiendly patch wobot.
Branch in linux-omap: omap-fixes
Initial commit ID (Likely to change): c9846244fa63adfa4930e35a1ee8bec02e0dd1b4
PatchWorks
http://patchwork.kernel.org/patch/84640/
Git (Likely to change, and takes a while to get mirrored)
http://git.kernel.org/?p=linux/kernel/git/tmlind/linux-omap-2.6.git;a=commit;h=c9846244fa63adfa4930e35a1ee8bec02e0dd1b4
PATCH 2/2 OMAP4: fix temporary hacks that break multi-omap PM
In reply to Kevin HilmanWhen building for multi-omap, and OMAP4 is enabled, CONFIG_ARCH_OMAP4 will be true and prevent included code from building/running for OMAP2/3 as well.
This problem exists in io.c where some hwmod/PM/SDRC init code is prevented from running even on OMAP2/3 when OMAP4 is included in a multi-OMAP build.
A quick glance suggests that this #ifndef is no longer needed in most of the cases. In the remaining cases, the function is wrapped with "if (cpu_is_omap24xx() || cpu_is_omap34xx())" which will be optimized out for OMAP4-only builds.
Note that this is only a short-term fix. Longer-term, OMAP4 needs to create init functions for SDRC and hwmod late-init.
--- arch/arm/mach-omap2/io.c | 13 ++++++------- 1 files changed, 6 insertions(+), 7 deletions(-)
arch/arm/mach-omap2/io.c +6 -7
@@ -309,7 +309,6 @@ void __init omap2_init_common_hw(struct omap_sdrc_params *sdrc_cs0, {pwrdm_init (powerdomains_omap);clkdm_init (clockdomains_omap, clkdm_autodeps); -#ifndef CONFIG_ARCH_OMAP4 /* FIXME: Remove this once the clkdev is ready */ if (cpu_is_omap242x ())omap2420_hwmod_init ();else if (cpu_is_omap243x ())
@@ -319,7 +318,6 @@ void __init omap2_init_common_hw(struct omap_sdrc_params *sdrc_cs0,omap2_mux_init ();/* The OPP tables have to be registered before a clk init */ omap_pm_if_early_init (mpu_opps, dsp_opps, l3_opps); -#endif if (cpu_is_omap2420 ())omap2420_clk_init ();
@@ -333,11 +331,12 @@ void __init omap2_init_common_hw(struct omap_sdrc_params *sdrc_cs0,pr_err ("Could not init clock framework - unknown CPU\n" );omap_serial_early_init (); -#ifndef CONFIG_ARCH_OMAP4 -omap_hwmod_late_init (); +if (cpu_is_omap24xx () ||cpu_is_omap34xx ())/* FIXME: OMAP4 */ +omap_hwmod_late_init ();omap_pm_if_init (); -omap2_sdrc_init (sdrc_cs0, sdrc_cs1); -_omap2_init_reprogram_sdrc (); -#endif +if (cpu_is_omap24xx () ||cpu_is_omap34xx ()) { +omap2_sdrc_init (sdrc_cs0, sdrc_cs1); +_omap2_init_reprogram_sdrc (); + }gpmc_init (); }
Signature · 1 line
1.7.0.2
Tony Lindgren
maintainerRe: [PATCH 2/2] OMAP4: fix temporary hacks that break multi-omap PM
Quoted 8 lines · Kevin Hilman
When building for multi-omap, and OMAP4 is enabled,
CONFIG_ARCH_OMAP4will be true and prevent included code from building/running for OMAP2/3 as well.This problem exists in
io.cwhere some hwmod/PM/SDRC init code is prevented from running even on OMAP2/3 when OMAP4 is included in a multi-OMAP build.
Quoted 7 lines
A quick glance suggests that this #ifndef is no longer needed in most of the cases. In the remaining cases, the function is wrapped with "if (
cpu_is_omap24xx()||cpu_is_omap34xx())" which will be optimized out for OMAP4-only builds.Note that this is only a short-term fix. Longer-term, OMAP4 needs to create init functions for SDRC and hwmod late-init.
Good to get rid of the temporary hacks. Santosh, maybe grep to see if there are other hacks left that can be now removed?
Regards,
Tony
Re: [PATCH 2/2] OMAP4: fix temporary hacks that break multi-omap PM
Quoted 1 line · Tony Lindgren
* Kevin Hilman <[email protected]> [100310 09:16]:
[...]
Quoted 5 lines
Quoted 2 lines
Note that this is only a short-term fix. Longer-term, OMAP4 needs to create init functions for SDRC and hwmod late-init.
Good to get rid of the temporary hacks. Santosh, maybe grep to see if there are other hacks left that can be now removed?
Oops, I shoul've removed last comment about SDRC init. I think the only thing missing is an OMAP4 hwmod init, which is work in progress with the auto-generated hwmods.
Kevin
RE: [PATCH 2/2] OMAP4: fix temporary hacks that break multi-omap PM
In reply to Tony LindgrenQuoted 21 lines
-----Original Message-----
From: Tony Lindgren [mailto:[email protected]]
Sent: Thursday, March 11, 2010 3:13 AM
To: Kevin Hilman
Cc: [email protected]; Shilimkar, Santosh
Subject: Re: [PATCH 2/2] OMAP4: fix temporary hacks that break multi-omap PMQuoted 13 lines · Kevin Hilman
When building for multi-omap, and OMAP4 is enabled,
CONFIG_ARCH_OMAP4will be true and prevent included code from building/running for OMAP2/3 as well.This problem exists in
io.cwhere some hwmod/PM/SDRC init code is prevented from running even on OMAP2/3 when OMAP4 is included in a multi-OMAP build.A quick glance suggests that this #ifndef is no longer needed in most of the cases. In the remaining cases, the function is wrapped with "if (
cpu_is_omap24xx()||cpu_is_omap34xx())" which will be optimized out for OMAP4-only builds.
Quoted 6 lines
Quoted 2 lines
Note that this is only a short-term fix. Longer-term, OMAP4 needs to create init functions for SDRC and hwmod late-init.
Good to get rid of the temporary hacks. Santosh, maybe grep to see if there are other hacks left that can be now removed?
Tony,
We have almost cleaned up all the hacks in platform code except the one around ongoing hwmod work.
Will check if I have missed anything.
Regards,
Santosh
Tony Lindgren
maintainerPATCH 2/2 OMAP4: fix temporary hacks that break multi-omap PM
In reply to Kevin HilmanThis patch has been applied to the linux-omap
by youw fwiendly patch wobot.
Branch in linux-omap: omap-fixes
Initial commit ID (Likely to change): f5564bf423cb747c82101d397655216c9d98682f
PatchWorks
http://patchwork.kernel.org/patch/84641/
Git (Likely to change, and takes a while to get mirrored)
http://git.kernel.org/?p=linux/kernel/git/tmlind/linux-omap-2.6.git;a=commit;h=f5564bf423cb747c82101d397655216c9d98682f