Re: [tip: perf/urgent] perf/x86: Check data address for IBS software filter

[Peter Zijlstra <peterz@infradead.org>]

On Mon, Mar 17, 2025 at 09:15:05AM -0000, tip-bot2 for Namhyung Kim wrote:
> The following commit has been merged into the perf/urgent branch of tip:
> 
> Commit-ID:     b0be17d8108bf3448a58be319d085155a128cf3a
> Gitweb:        https://git.kernel.org/tip/b0be17d8108bf3448a58be319d085155a128cf3a
> Author:        Namhyung Kim <namhyung@kernel.org>
> AuthorDate:    Mon, 17 Mar 2025 01:10:58 -07:00
> Committer:     Ingo Molnar <mingo@kernel.org>
> CommitterDate: Mon, 17 Mar 2025 10:04:31 +01:00
> 
> perf/x86: Check data address for IBS software filter
> 
> The IBS software filter is filtering kernel samples for regular users in
> PMI handler.  It checks the instruction address in the IBS register to
> determine if it was in the kernel mode or not.
> 
> But it turns out that it's possible to report a kernel data address even
> if the instruction address belongs to the user space.  Matteo Rizzo
> found that when an instruction raises an exception, IBS can report some
> kernel data address like IDT while holding the faulting instruction's
> RIP.  To prevent an information leak, it should double check if the data
> address in PERF_SAMPLE_DATA is in the kernel space as well.
> 
> Suggested-by: Matteo Rizzo <matteorizzo@google.com>
> Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> Signed-off-by: Ingo Molnar <mingo@kernel.org>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Link: https://lore.kernel.org/r/20250317081058.1794729-1-namhyung@kernel.org
> ---
>  arch/x86/events/amd/ibs.c | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/arch/x86/events/amd/ibs.c b/arch/x86/events/amd/ibs.c
> index e7a8b87..24985c7 100644
> --- a/arch/x86/events/amd/ibs.c
> +++ b/arch/x86/events/amd/ibs.c
> @@ -1147,6 +1147,13 @@ fail:
>  	if (perf_ibs == &perf_ibs_op)
>  		perf_ibs_parse_ld_st_data(event->attr.sample_type, &ibs_data, &data);
>  
> +	if ((event->attr.config2 & IBS_SW_FILTER_MASK) &&
> +	    (event->attr.sample_type & PERF_SAMPLE_ADDR) &&
> +	    event->attr.exclude_kernel && !access_ok(data.addr)) {

If only you'd looked at all the other filter code :/ everybody uses
kernel_ip() helper for this, not access_ok().

> +		throttle = perf_event_account_interrupt(event);
> +		goto out;
> +	}
> +
>  	/*
>  	 * rip recorded by IbsOpRip will not be consistent with rsp and rbp
>  	 * recorded as part of interrupt regs. Thus we need to use rip from