Re: [PATCH] Un-inline spinlocks on ppc64

[Keith Owens]

On Mon, 10 May 2004 10:11:33 +1000, 
Paul Mackerras <paulus@xxxxxxxxx> wrote:
>The one problem with out-of-line spinlock routines is that lock
>contention will show up in profiles in the spin_lock etc. routines
>rather than in the callers, as it does with inline spinlocks.  I have
>added a CONFIG_INLINE_SPINLOCKS config option for people that want to
>do profiling.  In the longer term, Anton is talking about teaching the
>profiling code to attribute samples in the spin lock routines to the
>routine's caller.

If it is any help, SGI patched the ia64 perfmon code to cope with a 2.4
version of the out of line spinlock patch.  The ia64 code uses a
non-standard calling sequence to work around a bug in gcc prior to 3.4,
r28 is the return address for the out of line code.

pfm_record_sample()
{
	...
#ifdef CONFIG_SMP
        extern char ia64_spinlock_contention[], ia64_spinlock_contention_end[];
#endif
	...
        h->ip   = regs ? regs->cr_iip | ((regs->cr_ipsr >> 41) & 0x3): 0x0UL;
#ifdef CONFIG_SMP
        if (h->ip >= (unsigned long)ia64_spinlock_contention &&
                        h->ip < (unsigned long)ia64_spinlock_contention_end)
                h->ip = regs->r28;
#endif
	...
}

-
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/