From: Kyle Zeng <kylebot@openai.com>
To: linux-perf-users@vger.kernel.org
Cc: linux-kernel@vger.kernel.org, peterz@infradead.org,
mingo@redhat.com, acme@kernel.org, namhyung@kernel.org,
outbounddisclosures@openai.com, Kyle Zeng <kylebot@openai.com>
Subject: [PATCH] perf/core: Require kernel access for page-size samples
Date: Tue, 6 Oct 2026 15:50:33 -0700 [thread overview]
Message-ID: <20261006225033.53296-1-kylebot@openai.com> (raw)
PERF_SAMPLE_CODE_PAGE_SIZE and PERF_SAMPLE_DATA_PAGE_SIZE translate a
sampled virtual address using the current task's page tables, including
kernel mappings. exclude_kernel only filters the execution mode of the
sample; it does not establish that the address being translated belongs
to userspace.
For example, an unprivileged tracer can put a kernel address in a
tracee's saved IP at a syscall-entry stop. A following syscall perf
tracepoint still sees user-mode registers and reports the page size at
that address. The distinction between unmapped and mapped addresses
can disclose the randomized kernel image base even at
perf_event_paranoid=2.
Require perf_allow_kernel() for both page-size sample types, as is
already done for PERF_SAMPLE_PHYS_ADDR. This applies the existing perf
and LSM authorization policy before the event is created, independently
of the source or representation of the sampled address. Authorized
page-size sampling and events that do not request these translation
fields are unchanged.
Fixes: 8d97e71811aa ("perf/core: Add PERF_SAMPLE_DATA_PAGE_SIZE")
Fixes: 995f088efebe ("perf/core: Add support for PERF_SAMPLE_CODE_PAGE_SIZE")
Assisted-by: Codex:gpt-6-astra
Signed-off-by: Kyle Zeng <kylebot@openai.com>
---
kernel/events/core.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/kernel/events/core.c b/kernel/events/core.c
index db7b76d6b68aa55dbe95e594d98a47230d12e979..97e5db887bd0a6e45437674670bc38efc6d7d9be 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -13967,8 +13967,13 @@ SYSCALL_DEFINE5(perf_event_open,
return -EINVAL;
}
- /* Only privileged users can get physical addresses */
- if ((attr.sample_type & PERF_SAMPLE_PHYS_ADDR)) {
+ /*
+ * Address translation can disclose kernel mappings even when the
+ * sampled registers are from user mode.
+ */
+ if (attr.sample_type & (PERF_SAMPLE_PHYS_ADDR |
+ PERF_SAMPLE_DATA_PAGE_SIZE |
+ PERF_SAMPLE_CODE_PAGE_SIZE)) {
err = perf_allow_kernel();
if (err)
return err;
base-commit: fd179f8a05be3ccae366b9b96e176b51fbe54aab
reply other threads:[~2026-10-06 22:50 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20261006225033.53296-1-kylebot@openai.com \
--to=kylebot@openai.com \
--cc=acme@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=namhyung@kernel.org \
--cc=outbounddisclosures@openai.com \
--cc=peterz@infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®