Re: [PATCH v2] powerpc 32: Provides VIRT_CPU_ACCOUNTING

From: Benjamin Herrenschmidt
Date: Wed Apr 30 2014 - 01:16:52 EST


On Mon, 2014-04-07 at 09:31 +0200, Christophe Leroy wrote:
> This patch provides VIRT_CPU_ACCOUTING to PPC32 architecture.
> Unlike PPC64, PPC32 doesn't use the PACA convention. Therefore the
> implementation is taken from the IA64 architecture.
> It is based on additional information added to the Task Info structure.
>
> Signed-off-by: Christophe Leroy <christophe.leroy@xxxxxx>

Scott, Can you review/ack this (or get somebody to) ?

It looks like a great idea but I really don't have the bandwidth to
review in detail and test right now.

(Adding Alister as well who maintains our 4xx 32-bit stuff nowadays).

Cheers,
Ben.

> Index: b/arch/powerpc/Kconfig
> ===================================================================
> --- a/arch/powerpc/Kconfig (revision 5607)
> +++ b/arch/powerpc/Kconfig (revision 5611)
> @@ -138,6 +138,7 @@
> select OLD_SIGSUSPEND
> select OLD_SIGACTION if PPC32
> select HAVE_DEBUG_STACKOVERFLOW
> + select HAVE_VIRT_CPU_ACCOUNTING
>
> config EARLY_PRINTK
> bool
> Index: a/arch/powerpc/kernel/time.c
> ===================================================================
> --- a/arch/powerpc/kernel/time.c (revision 5607)
> +++ b/arch/powerpc/kernel/time.c (revision 5611)
> @@ -162,7 +162,9 @@
>
> cputime_t cputime_one_jiffy;
>
> +#ifdef CONFIG_PPC_SPLPAR
> void (*dtl_consumer)(struct dtl_entry *, u64);
> +#endif
>
> static void calc_cputime_factors(void)
> {
> @@ -178,6 +180,7 @@
> __cputime_clockt_factor = res.result_low;
> }
>
> +#ifdef CONFIG_PPC64
> /*
> * Read the SPURR on systems that have it, otherwise the PURR,
> * or if that doesn't exist return the timebase value passed in.
> @@ -190,6 +193,7 @@
> return mfspr(SPRN_PURR);
> return tb;
> }
> +#endif
>
> #ifdef CONFIG_PPC_SPLPAR
>
> @@ -291,6 +295,7 @@
> * Account time for a transition between system, hard irq
> * or soft irq state.
> */
> +#ifdef CONFIG_PPC64
> static u64 vtime_delta(struct task_struct *tsk,
> u64 *sys_scaled, u64 *stolen)
> {
> @@ -377,7 +382,70 @@
> get_paca()->utime_sspurr = 0;
> account_user_time(tsk, utime, utimescaled);
> }
> +#else
>
> +void vtime_account_user(struct task_struct *tsk)
> +{
> + cputime_t delta_utime;
> + struct thread_info *ti = task_thread_info(tsk);
> +
> + if (ti->ac_utime) {
> + delta_utime = ti->ac_utime;
> + account_user_time(tsk, delta_utime, delta_utime);
> + ti->ac_utime = 0;
> + }
> +}
> +
> +/*
> + * Called from the context switch with interrupts disabled, to charge all
> + * accumulated times to the current process, and to prepare accounting on
> + * the next process.
> + */
> +void arch_vtime_task_switch(struct task_struct *prev)
> +{
> + struct thread_info *pi = task_thread_info(prev);
> + struct thread_info *ni = task_thread_info(current);
> +
> + ni->ac_stamp = pi->ac_stamp;
> + ni->ac_stime = ni->ac_utime = 0;
> +}
> +
> +/*
> + * Account time for a transition between system, hard irq or soft irq state.
> + * Note that this function is called with interrupts enabled.
> + */
> +static cputime_t vtime_delta(struct task_struct *tsk)
> +{
> + struct thread_info *ti = task_thread_info(tsk);
> + __u32 delta_stime;
> + __u32 now;
> +
> + WARN_ON_ONCE(!irqs_disabled());
> +
> + now = mftbl();
> +
> + delta_stime = ti->ac_stime + (now - ti->ac_stamp);
> + ti->ac_stime = 0;
> + ti->ac_stamp = now;
> +
> + return (cputime_t)delta_stime;
> +}
> +
> +void vtime_account_system(struct task_struct *tsk)
> +{
> + cputime_t delta = vtime_delta(tsk);
> +
> + account_system_time(tsk, 0, delta, delta);
> +}
> +EXPORT_SYMBOL_GPL(vtime_account_system);
> +
> +void vtime_account_idle(struct task_struct *tsk)
> +{
> + account_idle_time(vtime_delta(tsk));
> +}
> +
> +#endif
> +
> #else /* ! CONFIG_VIRT_CPU_ACCOUNTING_NATIVE */
> #define calc_cputime_factors()
> #endif
> @@ -871,6 +939,8 @@
> ppc_proc_freq / 1000000, ppc_proc_freq % 1000000);
> }
>
> + mttbl(0);
> + mttbu(0);
> tb_ticks_per_jiffy = ppc_tb_freq / HZ;
> tb_ticks_per_sec = ppc_tb_freq;
> tb_ticks_per_usec = ppc_tb_freq / 1000000;
> Index: b/arch/powerpc/kernel/entry_32.S
> ===================================================================
> --- a/arch/powerpc/kernel/entry_32.S (revision 5607)
> +++ b/arch/powerpc/kernel/entry_32.S (revision 5611)
> @@ -177,6 +177,12 @@
> addi r12,r12,-1
> stw r12,4(r11)
> #endif
> +#ifdef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
> + CURRENT_THREAD_INFO(r9, r1)
> + tophys(r9, r9)
> + ACCOUNT_CPU_USER_ENTRY(r9, r11, r12)
> +#endif
> +
> b 3f
>
> 2: /* if from kernel, check interrupted DOZE/NAP mode and
> @@ -406,6 +412,13 @@
> lwarx r7,0,r1
> END_FTR_SECTION_IFSET(CPU_FTR_NEED_PAIRED_STWCX)
> stwcx. r0,0,r1 /* to clear the reservation */
> +#ifdef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
> + andi. r4,r8,MSR_PR
> + beq 3f
> + CURRENT_THREAD_INFO(r4, r1)
> + ACCOUNT_CPU_USER_EXIT(r4, r5, r7)
> +3:
> +#endif
> lwz r4,_LINK(r1)
> lwz r5,_CCR(r1)
> mtlr r4
> @@ -841,6 +854,10 @@
> andis. r10,r0,DBCR0_IDM@h
> bnel- load_dbcr0
> #endif
> +#ifdef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
> + CURRENT_THREAD_INFO(r9, r1)
> + ACCOUNT_CPU_USER_EXIT(r9, r10, r11)
> +#endif
>
> b restore
>
> Index: b/arch/powerpc/kernel/asm-offsets.c
> ===================================================================
> --- a/arch/powerpc/kernel/asm-offsets.c (revision 5607)
> +++ b/arch/powerpc/kernel/asm-offsets.c (revision 5611)
> @@ -167,6 +167,12 @@
> DEFINE(TI_PREEMPT, offsetof(struct thread_info, preempt_count));
> DEFINE(TI_TASK, offsetof(struct thread_info, task));
> DEFINE(TI_CPU, offsetof(struct thread_info, cpu));
> +#if defined(CONFIG_VIRT_CPU_ACCOUNTING_NATIVE) && defined(CONFIG_PPC32)
> + DEFINE(TI_AC_STAMP, offsetof(struct thread_info, ac_stamp));
> + DEFINE(TI_AC_LEAVE, offsetof(struct thread_info, ac_leave));
> + DEFINE(TI_AC_STIME, offsetof(struct thread_info, ac_stime));
> + DEFINE(TI_AC_UTIME, offsetof(struct thread_info, ac_utime));
> +#endif
>
> #ifdef CONFIG_PPC64
> DEFINE(DCACHEL1LINESIZE, offsetof(struct ppc64_caches, dline_size));
> Index: b/arch/powerpc/include/asm/thread_info.h
> ===================================================================
> --- a/arch/powerpc/include/asm/thread_info.h (revision 5607)
> +++ b/arch/powerpc/include/asm/thread_info.h (revision 5611)
> @@ -43,6 +43,12 @@
> int cpu; /* cpu we're on */
> int preempt_count; /* 0 => preemptable,
> <0 => BUG */
> +#if defined(CONFIG_VIRT_CPU_ACCOUNTING_NATIVE) && defined(CONFIG_PPC32)
> + u32 ac_stamp;
> + u32 ac_leave;
> + u32 ac_stime;
> + u32 ac_utime;
> +#endif
> struct restart_block restart_block;
> unsigned long local_flags; /* private flags for thread */
>
> Index: b/arch/powerpc/include/asm/cputime.h
> ===================================================================
> --- a/arch/powerpc/include/asm/cputime.h (revision 5607)
> +++ b/arch/powerpc/include/asm/cputime.h (revision 5611)
> @@ -228,7 +228,11 @@
>
> #define cputime64_to_clock_t(ct) cputime_to_clock_t((cputime_t)(ct))
>
> +#ifdef CONFIG_PPC64
> static inline void arch_vtime_task_switch(struct task_struct *tsk) { }
> +#else
> +extern void arch_vtime_task_switch(struct task_struct *tsk);
> +#endif
>
> #endif /* __KERNEL__ */
> #endif /* CONFIG_VIRT_CPU_ACCOUNTING_NATIVE */
> Index: b/arch/powerpc/include/asm/ppc_asm.h
> ===================================================================
> --- a/arch/powerpc/include/asm/ppc_asm.h (revision 5607)
> +++ b/arch/powerpc/include/asm/ppc_asm.h (revision 5611)
> @@ -25,10 +25,16 @@
> */
>
> #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
> +#ifdef CONFIG_PPC64
> #define ACCOUNT_CPU_USER_ENTRY(ra, rb)
> #define ACCOUNT_CPU_USER_EXIT(ra, rb)
> +#else /* CONFIG_PPC64 */
> +#define ACCOUNT_CPU_USER_ENTRY(ti, ra, rb)
> +#define ACCOUNT_CPU_USER_EXIT(ti, ra, rb)
> +#endif /* CONFIG_PPC64 */
> #define ACCOUNT_STOLEN_TIME
> -#else
> +#else /* CONFIG_VIRT_CPU_ACCOUNTING_NATIVE */
> +#ifdef CONFIG_PPC64
> #define ACCOUNT_CPU_USER_ENTRY(ra, rb) \
> MFTB(ra); /* get timebase */ \
> ld rb,PACA_STARTTIME_USER(r13); \
> @@ -68,7 +74,27 @@
> #define ACCOUNT_STOLEN_TIME
>
> #endif /* CONFIG_PPC_SPLPAR */
> +#else /* CONFIG_PPC64 */
> +#define ACCOUNT_CPU_USER_ENTRY(ti, ra, rb) \
> + MFTB(ra); \
> + lwz rb, TI_AC_LEAVE(ti); \
> + stw ra, TI_AC_STAMP(ti); /* AC_STAMP = NOW */ \
> + subf rb, rb, ra; /* R = NOW - AC_LEAVE */ \
> + lwz ra, TI_AC_UTIME(ti); \
> + add ra, rb, ra; /* AC_UTIME += R */ \
> + stw ra, TI_AC_UTIME(ti); \
>
> +#define ACCOUNT_CPU_USER_EXIT(ti, ra, rb) \
> + MFTB(ra); \
> + lwz rb, TI_AC_STAMP(ti); \
> + stw ra, TI_AC_LEAVE(ti); \
> + subf rb, rb, ra; /* R = NOW - AC_STAMP */ \
> + lwz ra, TI_AC_STIME(ti); \
> + add ra, rb, ra; /* AC_STIME += R */ \
> + stw ra, TI_AC_STIME(ti); \
> +
> +#endif /* CONFIG_PPC64 */
> +
> #endif /* CONFIG_VIRT_CPU_ACCOUNTING_NATIVE */
>
> /*
> Index: b/arch/powerpc/platforms/Kconfig.cputype
> ===================================================================
> --- a/arch/powerpc/platforms/Kconfig.cputype (revision 5607)
> +++ b/arch/powerpc/platforms/Kconfig.cputype (revision 5611)
> @@ -1,7 +1,6 @@
> config PPC64
> bool "64-bit kernel"
> default n
> - select HAVE_VIRT_CPU_ACCOUNTING
> help
> This option selects whether a 32-bit or a 64-bit kernel
> will be built.


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/