linux-kernel - Re: [PATCH] perf/x86/intel/lbr: fix branch type encoding

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <4a44b08c-ff30-3236-9e3e-775e4905561a@linux.intel.com>
Date:   Thu, 11 Aug 2022 22:21:43 +0200
From:   Andi Kleen <ak@...ux.intel.com>
To:     Stephane Eranian <eranian@...gle.com>,
        "Liang, Kan" <kan.liang@...ux.intel.com>
Cc:     linux-kernel@...r.kernel.org, peterz@...radead.org,
        kan.liang@...el.com, acme@...hat.com, namhyung@...nel.org,
        irogers@...gle.com
Subject: Re: [PATCH] perf/x86/intel/lbr: fix branch type encoding

On 8/11/2022 5:33 PM, Stephane Eranian wrote:
>
> In fact, I don't see a case where you would not benefit from the branch type.
> Furthermore, not having the branch type DOES NOT save any space in the
> branch record (given we have a reserved field). So I think I prefer not having
> to specify yet another cmdline option to get the branch type. In fact, if you do
> not pass the option, then perf report -D reports some bogus branch types, i.e.,
> not all entries have empty types.

Perhaps Peter remember the details, but there was some ABI compatibility 
issue that motivated the extra save type bitmap.

Also there's another reason that if you don't need them and the hardware 
doesn't support it they add run time overhead because the branches need 
to be decoded in the PMI.

But I agree it would be useful to make them part of -b default, as long 
as there is some way to turn it off again when not needed. The extra 
option is kind of a pain.

-Andi