Re: [PATCH v2 25/60] scsi: qla2xxx: Use ring-slot helpers in __qla2x00_alloc_iocbs

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Standard `memset()` is used on an `__iomem` mapped request ring pointer for QLAFX00 adapters, which can cause unaligned access faults or bus errors on architectures like ARM64.
--

commit 9791936eaf3c9274e9174a4c671ecce5a7c3c126
Author: Nilesh Javali <njavali@marvell.com>

scsi: qla2xxx: Use ring-slot helpers in __qla2x00_alloc_iocbs

This commit refactors __qla2x00_alloc_iocbs() to use qla_req_ring_slot() and
qla_req_entry_size() helpers instead of open-coding the ring pointer selection
and entry size based on the adapter type.

> diff --git a/drivers/scsi/qla2xxx/qla_iocb.c b/drivers/scsi/qla2xxx/qla_iocb.c
> index d16570a6a0724..e2b64029a5743 100644
> --- a/drivers/scsi/qla2xxx/qla_iocb.c
> +++ b/drivers/scsi/qla2xxx/qla_iocb.c
[ ... ]
> @@ -2469,13 +2468,8 @@ __qla2x00_alloc_iocbs(struct qla_qpair *qpair, srb_t *sp)
>  	 * layout-compatible once we return the pkt as request_t *.
>  	 */
>  	req->cnt -= req_cnt;
> -	if (IS_QLA29XX(ha)) {
> -		pkt = (request_t *)req->ring_ext_ptr;
> -		memset(pkt, 0, REQUEST_ENTRY_SIZE_EXT);
> -	} else {
> -		pkt = req->ring_ptr;
> -		memset(pkt, 0, REQUEST_ENTRY_SIZE);
> -	}
> +	pkt = qla_req_ring_slot(ha, req);
> +	memset(pkt, 0, qla_req_entry_size(ha));
>  	if (IS_QLAFX00(ha)) {
>  		wrt_reg_byte((u8 __force __iomem *)&pkt->entry_count, req_cnt);

[Severity: High]
This is a pre-existing issue, but does using standard memset() on the request
ring slot pointer cause problems for QLAFX00 adapters?

For QLAFX00 adapters, req->ring is mapped I/O memory, which is why the
register-mapped writes below require __iomem casts. On architectures like
ARM64, calling standard memset() directly on device memory can utilize
unaligned or optimized instructions that are invalid for I/O memory, leading
to a synchronous external abort and a kernel panic. 

Should this be using an I/O specific memset, or could the memset be
skipped entirely for QLAFX00 device memory?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260612095333.1666592-1-njavali@marvell.com?part=25