Re: [PATCH v6 06/10] ACPI: APEI: GHES: move CXL CPER helpers

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:

Pre-existing issues:
- [High] Inconsistent spinlock usage between process context and interrupt context for `cxl_cper_work_lock` and `cxl_cper_prot_err_work_lock` causes a potential deadlock.
- [High] Unconditional fixed-size copy leads to out-of-bounds read if the ACPI payload is short.
- [High] Lack of bounds checking for DVSEC length leads to out-of-bounds read in CXL protocol error handling.
--

commit bb34e6b6044e94825caae9bb7ad029942aef5c28
Author: Ahmed Tiba <ahmed.tiba@arm.com>

ACPI: APEI: GHES: move CXL CPER helpers

Move the CXL CPER handling paths out of ghes.c and into ghes_cper.c so the
helpers can be reused. The code is moved as-is, with the public
prototypes updated so GHES keeps calling into the new translation unit.

> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
[ ... ]
> diff --git a/drivers/acpi/apei/ghes_cper.c b/drivers/acpi/apei/ghes_cper.c
> index b26943eafd792..66bf1af4db00b 100644
> --- a/drivers/acpi/apei/ghes_cper.c
> +++ b/drivers/acpi/apei/ghes_cper.c
[ ... ]
> @@ -336,6 +339,138 @@ void ghes_defer_non_standard_event(struct acpi_hest_generic_data *gdata,
[ ... ]
> +void cxl_cper_post_prot_err(struct cxl_cper_sec_prot_err *prot_err,
> +			    int severity)
> +{
> +#ifdef CONFIG_ACPI_APEI_PCIEAER
> +	struct cxl_cper_prot_err_work_data wd;
> +
> +	if (cxl_cper_sec_prot_err_valid(prot_err))
> +		return;
> +
> +	guard(spinlock_irqsave)(&cxl_cper_prot_err_work_lock);
> +
> +	if (!cxl_cper_prot_err_work)
> +		return;
> +
> +	if (cxl_cper_setup_prot_err_work_data(&wd, prot_err, severity))

[Severity: High]
This is a pre-existing issue, but could this setup step read out of bounds?

The cxl_cper_setup_prot_err_work_data() function uses the firmware-provided
dvsec_len to compute a pointer for copying the RAS capability registers:

ghes_helpers.c:cxl_cper_setup_prot_err_work_data() {
        ...
        cap_start = dvsec_start + prot_err->dvsec_len;
        memcpy(&wd->ras_cap, cap_start, sizeof(wd->ras_cap));

Since cxl_cper_post_prot_err() and its callers do not validate that
dvsec_len falls within the maximum bounds of the ACPI payload
(gdata->error_data_length), could a malformed firmware record cause cap_start
to point beyond the valid buffer?

> +		return;
> +
> +	if (!kfifo_put(&cxl_cper_prot_err_fifo, wd)) {
> +		pr_err_ratelimited("CXL CPER kfifo overflow\n");
> +		return;
> +	}
> +
> +	schedule_work(cxl_cper_prot_err_work);
> +#endif
> +}
> +
> +int cxl_cper_register_prot_err_work(struct work_struct *work)
> +{
> +	if (cxl_cper_prot_err_work)
> +		return -EINVAL;
> +
> +	guard(spinlock)(&cxl_cper_prot_err_work_lock);

[Severity: High]
This isn't a bug introduced by this patch, but is there a potential deadlock
between process context and interrupt context here?

The registration function acquires the lock with guard(spinlock), leaving
local interrupts enabled.

If a hardware interrupt occurs while this lock is held, and the interrupt
handler calls cxl_cper_post_prot_err() which attempts to acquire the same
lock using guard(spinlock_irqsave):

cxl_cper_post_prot_err() {
        ...
        guard(spinlock_irqsave)(&cxl_cper_prot_err_work_lock);

Could this cause the interrupt handler to spin forever on the local CPU?

> +	cxl_cper_prot_err_work = work;
> +	return 0;
> +}
[ ... ]
> +void cxl_cper_post_event(enum cxl_event_type event_type,
> +			 struct cxl_cper_event_rec *rec)
> +{
> +	struct cxl_cper_work_data wd;
> +
> +	if (rec->hdr.length <= sizeof(rec->hdr) ||
> +	    rec->hdr.length > sizeof(*rec)) {
> +		pr_err(FW_WARN "CXL CPER Invalid section length (%u)\n",
> +		       rec->hdr.length);
> +		return;
> +	}
> +
> +	if (!(rec->hdr.validation_bits & CPER_CXL_COMP_EVENT_LOG_VALID)) {
> +		pr_err(FW_WARN "CXL CPER invalid event\n");
> +		return;
> +	}
> +
> +	guard(spinlock_irqsave)(&cxl_cper_work_lock);
> +
> +	if (!cxl_cper_work)
> +		return;
> +
> +	wd.event_type = event_type;
> +	memcpy(&wd.rec, rec, sizeof(wd.rec));

[Severity: High]
This is a pre-existing issue, but can this copy read out of bounds if the
ACPI payload is unexpectedly short?

The memcpy unconditionally copies sizeof(wd.rec) bytes from the
firmware-provided payload.

Since gdata->error_data_length is never verified to be at least the size
of the record before being passed to this function by
ghes_cper_handle_status(), would a malformed payload smaller than the
struct cause an uncontrolled read of kernel memory?

> +
> +	if (!kfifo_put(&cxl_cper_fifo, wd)) {
[ ... ]

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-topics-ahmtib01-ras_ffh_arm_internal_review-v6-0-91f725174aa0@arm.com?part=6