Arto Merilainen <amerilainen@xxxxxxxxxx> writes:
> On 28.7.2025 16.51, Aneesh Kumar K.V (Arm) wrote:
>
>> +static int pci_res_to_pdev_addr(struct rmi_pdev_addr_range *pdev_addr,
>> + unsigned int naddr, struct resource *res,
>> + unsigned int nres)
>> +{
>> + int i, j;
>> +
>> + for (i = 0, j = 0; i < naddr && j < nres; j++) {
>> + if (res[j].flags & IORESOURCE_MEM) {
>> + pdev_addr[i].base = res[j].start;
>> + pdev_addr[i].top = res[j].end;
>
> I think there is an off-by-one bug in res[j].end. As per the RMM
> specification the base address is inclusive and the top address is
> exclusive. Both res[j].start and res[j].end are inclusive, and hence
> res[j].end seems wrong.
>
>> + /* use the rid and MMIO resources of the epdev */
>> + params->rid_top = params->rid_base = rid;
>
> Similar issue here. As per the specification the rid_base is inclusive
> and the rid_top exclusive.
>
Thanks for the review comments. I'll update the patch with the suggested changes.
-aneesh
