| Message ID | 1371501264-12546-1-git-send-email-paul.gortmaker@windriver.com (mailing list archive) |
|---|---|
| State | Awaiting Upstream |
| Headers | show |
On Mon, 2013-06-17 at 16:34 -0400, Paul Gortmaker wrote: > The __cpuinit type of throwaway sections might have made sense > some time ago when RAM was more constrained, but now the savings > do not offset the cost and complications. For example, the fix in > commit 5e427ec2d0 ("x86: Fix bit corruption at CPU resume time") > is a good example of the nasty type of bugs that can be created > with improper use of the various __init prefixes. > > After a discussion on LKML[1] it was decided that cpuinit should go > the way of devinit and be phased out. Once all the users are gone, > we can then finally remove the macros themselves from linux/init.h. Fine by us ... I'd recommend just rolling this into one patch for all the cpuinit tags and sending it via the trivial tree (which goes to linux-next) that way we'll know if we do anything to compromise it. James -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 13-06-17 06:47 PM, James Bottomley wrote: > On Mon, 2013-06-17 at 16:34 -0400, Paul Gortmaker wrote: >> The __cpuinit type of throwaway sections might have made sense >> some time ago when RAM was more constrained, but now the savings >> do not offset the cost and complications. For example, the fix in >> commit 5e427ec2d0 ("x86: Fix bit corruption at CPU resume time") >> is a good example of the nasty type of bugs that can be created >> with improper use of the various __init prefixes. >> >> After a discussion on LKML[1] it was decided that cpuinit should go >> the way of devinit and be phased out. Once all the users are gone, >> we can then finally remove the macros themselves from linux/init.h. > > Fine by us ... I'd recommend just rolling this into one patch for all > the cpuinit tags and sending it via the trivial tree (which goes to > linux-next) that way we'll know if we do anything to compromise it. Thanks James, I'll add an ack'd by for you and keep it with the arch agnostic chunks. I do want it to go by the arch maintainers 1st though, and give them 1st right of refusal on carrying the change; e.g. for MIPS, Ralf wanted to carry the patch due to the level of tree activity still pending there. And for sparc, it touches some trampoline code, so I want to be sure I don't inflict a silent boot death on the SPARC folks. It will get in linux-next though -- I'll be asking Stephen for that shortly. P. -- > > James > > -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" 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/arch/parisc/kernel/firmware.c b/arch/parisc/kernel/firmware.c index f65fa48..2239590 100644 --- a/arch/parisc/kernel/firmware.c +++ b/arch/parisc/kernel/firmware.c @@ -150,7 +150,7 @@ static void convert_to_wide(unsigned long *addr) } #ifdef CONFIG_64BIT -void __cpuinit set_firmware_width_unlocked(void) +void set_firmware_width_unlocked(void) { int ret; @@ -167,7 +167,7 @@ void __cpuinit set_firmware_width_unlocked(void) * This function must be called before any pdc_* function that uses the * convert_to_wide function. */ -void __cpuinit set_firmware_width(void) +void set_firmware_width(void) { unsigned long flags; spin_lock_irqsave(&pdc_lock, flags); @@ -175,11 +175,13 @@ void __cpuinit set_firmware_width(void) spin_unlock_irqrestore(&pdc_lock, flags); } #else -void __cpuinit set_firmware_width_unlocked(void) { +void set_firmware_width_unlocked(void) +{ return; } -void __cpuinit set_firmware_width(void) { +void set_firmware_width(void) +{ return; } #endif /*CONFIG_64BIT*/ @@ -301,7 +303,7 @@ int pdc_chassis_warn(unsigned long *warn) return retval; } -int __cpuinit pdc_coproc_cfg_unlocked(struct pdc_coproc_cfg *pdc_coproc_info) +int pdc_coproc_cfg_unlocked(struct pdc_coproc_cfg *pdc_coproc_info) { int ret; @@ -322,7 +324,7 @@ int __cpuinit pdc_coproc_cfg_unlocked(struct pdc_coproc_cfg *pdc_coproc_info) * This PDC call returns the presence and status of all the coprocessors * attached to the processor. */ -int __cpuinit pdc_coproc_cfg(struct pdc_coproc_cfg *pdc_coproc_info) +int pdc_coproc_cfg(struct pdc_coproc_cfg *pdc_coproc_info) { int ret; unsigned long flags; diff --git a/arch/parisc/kernel/hardware.c b/arch/parisc/kernel/hardware.c index 9e2d2e4..6f4bbb8 100644 --- a/arch/parisc/kernel/hardware.c +++ b/arch/parisc/kernel/hardware.c @@ -1366,7 +1366,7 @@ const char *parisc_hardware_description(struct parisc_device_id *id) /* Interpret hversion (ret[0]) from PDC_MODEL(4)/PDC_MODEL_INFO(0) */ -enum cpu_type __cpuinit +enum cpu_type parisc_get_cpu_type(unsigned long hversion) { struct hp_cpu_type_mask *ptr; diff --git a/arch/parisc/kernel/processor.c b/arch/parisc/kernel/processor.c index c8fb61e..a128a94 100644 --- a/arch/parisc/kernel/processor.c +++ b/arch/parisc/kernel/processor.c @@ -73,7 +73,7 @@ extern int update_cr16_clocksource(void); /* from time.c */ * * FIXME: doesn't do much yet... */ -static void __cpuinit +static void init_percpu_prof(unsigned long cpunum) { struct cpuinfo_parisc *p; @@ -92,7 +92,7 @@ init_percpu_prof(unsigned long cpunum) * (return 1). If so, initialize the chip and tell other partners in crime * they have work to do. */ -static int __cpuinit processor_probe(struct parisc_device *dev) +static int processor_probe(struct parisc_device *dev) { unsigned long txn_addr; unsigned long cpuid; @@ -299,7 +299,7 @@ void __init collect_boot_cpu_data(void) * * o Enable CPU profiling hooks. */ -int __cpuinit init_per_cpu(int cpunum) +int init_per_cpu(int cpunum) { int ret; struct pdc_coproc_cfg coproc_cfg; diff --git a/arch/parisc/kernel/smp.c b/arch/parisc/kernel/smp.c index e3614fb..8a252f2 100644 --- a/arch/parisc/kernel/smp.c +++ b/arch/parisc/kernel/smp.c @@ -62,9 +62,9 @@ static int smp_debug_lvl = 0; volatile struct task_struct *smp_init_current_idle_task; /* track which CPU is booting */ -static volatile int cpu_now_booting __cpuinitdata; +static volatile int cpu_now_booting; -static int parisc_max_cpus __cpuinitdata = 1; +static int parisc_max_cpus = 1; static DEFINE_PER_CPU(spinlock_t, ipi_lock); @@ -328,7 +328,7 @@ void __init smp_callin(void) /* * Bring one cpu online. */ -int __cpuinit smp_boot_one_cpu(int cpuid, struct task_struct *idle) +int smp_boot_one_cpu(int cpuid, struct task_struct *idle) { const struct cpuinfo_parisc *p = &per_cpu(cpu_data, cpuid); long timeout; @@ -424,7 +424,7 @@ void smp_cpus_done(unsigned int cpu_max) } -int __cpuinit __cpu_up(unsigned int cpu, struct task_struct *tidle) +int __cpu_up(unsigned int cpu, struct task_struct *tidle) { if (cpu != 0 && cpu < parisc_max_cpus) smp_boot_one_cpu(cpu, tidle);
The __cpuinit type of throwaway sections might have made sense some time ago when RAM was more constrained, but now the savings do not offset the cost and complications. For example, the fix in commit 5e427ec2d0 ("x86: Fix bit corruption at CPU resume time") is a good example of the nasty type of bugs that can be created with improper use of the various __init prefixes. After a discussion on LKML[1] it was decided that cpuinit should go the way of devinit and be phased out. Once all the users are gone, we can then finally remove the macros themselves from linux/init.h. This removes all the parisc uses of the __cpuinit macros. [1] https://lkml.org/lkml/2013/5/20/589 Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> --- [This was generated against today's linux-next tree ; I'm assuming all pending parisc changes are in there currently.] arch/parisc/kernel/firmware.c | 14 ++++++++------ arch/parisc/kernel/hardware.c | 2 +- arch/parisc/kernel/processor.c | 6 +++--- arch/parisc/kernel/smp.c | 8 ++++---- 4 files changed, 16 insertions(+), 14 deletions(-)