Re: [PATCHv3] perf x86_64: Fix rsp register for system call fast path

[Frederic Weisbecker <fweisbec@gmail.com>]

On Wed, Oct 03, 2012 at 02:29:47PM +0200, Jiri Olsa wrote:
> +#ifdef CONFIG_X86_64
> +__weak void

Only annotate with __weak the default implementation you want to be
overriden. Here you want it to actually override the default __weak version.

> +arch_sample_regs_user_fixup(struct perf_regs_user *uregs, int kernel)
> +{
> +	/*
> +	 * If the perf event was triggered within the kernel code
> +	 * path, then it was either syscall or interrupt. While
> +	 * interrupt stores almost all user registers, the syscall
> +	 * fast path does not. At this point we can at least set
> +	 * rsp register right, which is crucial for dwarf unwind.
> +	 *
> +	 * The syscall_get_nr function returns -1 (orig_ax) for
> +	 * interrupt, and positive value for syscall.
> +	 *
> +	 * We have two race windows in here:
> +	 *
> +	 * 1) Few instructions from syscall entry until old_rsp is
> +	 *    set.
> +	 *
> +	 * 2) In syscall/interrupt path from entry until the orig_ax
> +	 *    is set.
> +	 *
> +	 * Above described race windows are fractional opposed to
> +	 * the syscall fast path, so we get much better results
> +	 * fixing rsp this way.
> +	 */
> +	if (kernel && (syscall_get_nr(current, uregs->regs) >= 0)) {
> +		/* Make a copy and link it to regs pointer. */
> +		memcpy(&uregs->regs_copy, uregs->regs, sizeof(*uregs->regs));
> +		uregs->regs = &uregs->regs_copy;
> +
> +		/* And fix the rsp. */
> +		uregs->regs->sp = this_cpu_read(old_rsp);
> +	}
> +}
> +#endif