From: sashiko-bot@kernel.org
To: "Anvesh Jain P" <anvesh.p@oss.qualcomm.com>
Cc: devicetree@vger.kernel.org, robh@kernel.org, conor+dt@kernel.org
Subject: Re: [PATCH v3 3/4] remoteproc: qcom: pas: Add coredump support for secondary processors
Date: Wed, 18 Nov 2026 16:03:06 +0000 [thread overview]
Message-ID: <sashiko-outbox-221408@kernel.org> (raw)
In-Reply-To: <20261118-oob-support-glymur-v3-3-2cbdfe208d56@oss.qualcomm.com>
Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [Medium] remoteproc: qcom: pas: duplicate dump segments in
qcom_pas_load() on recovery
- [Medium] remoteproc: qcom: pas: secondary coredump after masked
shutdown failure
--
commit 18ab6cc078bc90d357be63c7984c43a899ef428d
Author: Anvesh Jain P <anvesh.p@oss.qualcomm.com>
remoteproc: qcom: pas: Add coredump support for secondary processors
This commit adds coredump support for PAS remoteproc instances with a
secondary processor by registering dump segments for the secondary
firmware and mapping the secondary carveout during coredump collection.
> diff --git a/drivers/remoteproc/qcom_q6v5_pas.c b/drivers/remoteproc/qcom_q6v5_pas.c
> index 1c8c0c22c34d1..a167bd6072c94 100644
> --- a/drivers/remoteproc/qcom_q6v5_pas.c
> +++ b/drivers/remoteproc/qcom_q6v5_pas.c
[ ... ]
> @@ -336,6 +342,17 @@ static int qcom_pas_load(struct rproc *rproc, const struct firmware *fw)
> }
> }
>
> + if (pas->secondary_pas_id) {
> + ret = qcom_register_dump_segments(rproc, pas->secondary_firmware);
> + if (ret) {
> + dev_err(pas->dev,
> + "failed to register secondary dump segments: %d\n", ret);
> + rproc_coredump_cleanup(rproc);
> + qcom_pas_release_firmware(pas);
> + return ret;
> + }
> + }
> +
> return 0;
> }
[Severity: Medium]
Does registering secondary dump segments in qcom_pas_load() append
duplicate rproc_dump_segment entries to rproc->dump_segments on every
crash recovery?
Primary dump segments are registered in qcom_pas_parse_firmware(), which
runs only during rproc_fw_boot(). On crash recovery,
rproc_boot_recovery() runs rproc->ops->coredump() and then calls
rproc_start() -> rproc_load_segments() -> qcom_pas_load() without
clearing rproc->dump_segments, so each recovery allocates and appends
another copy of the secondary segments and produces corrupted coredump
ELFs with duplicate PT_LOAD headers. In addition, if
qcom_register_dump_segments() fails here during recovery,
rproc_coredump_cleanup(rproc) wipes the primary segments that were
registered at initial boot and will not be re-registered on later
recoveries.
[ ... ]
> @@ -748,11 +787,28 @@ static void qcom_pas_coredump(struct rproc *rproc)
> {
> struct qcom_pas *pas = rproc->priv;
>
> - pas->mem_region = qcom_pas_ctx_map(pas->pas_ctx);
> + if (list_empty(&rproc->dump_segments))
> + return;
> +
> + if (!pas->mem_region)
> + pas->mem_region = qcom_pas_ctx_map(pas->pas_ctx);
> if (!pas->mem_region)
> return;
>
> + if (qcom_pas_has_secondary(pas) && !pas->secondary_mem_region) {
> + pas->secondary_mem_region =
> + qcom_pas_ctx_map(pas->secondary_pas_ctx);
> + if (!pas->secondary_mem_region)
> + dev_err(pas->dev, "unable to map secondary memory for coredump\n");
> + }
> +
> rproc_coredump(rproc);
[Severity: Medium]
Can rproc_coredump() read pas->secondary_mem_region while the secondary
carveout is still locked by TrustZone or the hypervisor if shutting down
pas->secondary_pas_id failed?
rproc_boot_recovery() only skips coredump when rproc_stop() returns an
error, as fixed for primary and DTB PAS IDs in commit 9db31edf92dd
("remoteproc: qcom_q6v5_pas: Fix error masking in qcom_pas_stop()").
However, qcom_pas_stop() overwrites the return value of
qcom_pas_shutdown(pas->secondary_pas_id) with the subsequent shutdown
calls:
drivers/remoteproc/qcom_q6v5_pas.c:qcom_pas_stop() {
...
if (pas->secondary_pas_id) {
ret = qcom_pas_shutdown(pas->secondary_pas_id);
if (ret && pas->decrypt_shutdown)
ret = qcom_pas_shutdown_poll_decrypt(pas->secondary_pas_id);
if (ret)
dev_err(pas->dev, "failed to shutdown secondary: %d\n", ret);
...
}
...
ret = qcom_pas_shutdown(pas->pas_id);
}
If the secondary shutdown fails while the primary shutdown succeeds,
qcom_pas_stop() returns 0 and rproc_coredump() reads the still-protected
secondary carveout, triggering an XPU or stage-2 access violation.
--
Sashiko AI review · https://sashiko.dev/#/patchset/20261118-oob-support-glymur-v3-0-2cbdfe208d56@oss.qualcomm.com?part=3
next prev parent reply other threads:[~2026-11-18 16:03 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-18 15:36 [PATCH v3 0/4] Add Glymur OOBMSS support Anvesh Jain P
2026-11-18 15:36 ` [PATCH v3 1/4] dt-bindings: remoteproc: qcom: Add Glymur OOBMSS PAS Anvesh Jain P
2026-11-18 15:36 ` [PATCH v3 2/4] remoteproc: qcom: pas: Support secondary images Anvesh Jain P
2026-11-18 15:56 ` sashiko-bot
2026-11-18 15:36 ` [PATCH v3 3/4] remoteproc: qcom: pas: Add coredump support for secondary processors Anvesh Jain P
2026-11-18 16:03 ` sashiko-bot [this message]
2026-11-18 15:36 ` [PATCH v3 4/4] remoteproc: qcom: pas: Add Glymur OOBMSS PAS parameters Anvesh Jain P
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=sashiko-outbox-221408@kernel.org \
--to=sashiko-bot@kernel.org \
--cc=anvesh.p@oss.qualcomm.com \
--cc=conor+dt@kernel.org \
--cc=devicetree@vger.kernel.org \
--cc=robh@kernel.org \
--cc=sashiko-reviews@lists.linux.dev \
/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