* [PATCH 1/1] cxl/acpi: Fix using extra xormaps for address translation
@ 2026-01-15 13:16 Li Ming
2026-01-15 16:12 ` Dave Jiang
2026-01-15 17:56 ` Alison Schofield
0 siblings, 2 replies; 4+ messages in thread
From: Li Ming @ 2026-01-15 13:16 UTC (permalink / raw)
To: dave, jonathan.cameron, dave.jiang, alison.schofield,
vishal.l.verma, ira.weiny, dan.j.williams
Cc: linux-cxl, linux-kernel, Li Ming
CXL r3.1 Table 9-22 defines a fixed number of XOR maps for interleave
arithmetic based on the number of interleave ways, as reflected by
hbiw_to_nr_maps[].
The current implementation applies all XOR maps during address
translation, but it should use the spec-defined number of XOR maps
derived from the number of interleave ways.
Fixes: 3b2fedcd75e3 ("cxl: Restore XOR'd position bits during address translation")
Signed-off-by: Li Ming <ming.li@zohomail.com>
---
drivers/cxl/acpi.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/cxl/acpi.c b/drivers/cxl/acpi.c
index 77ac940e3013..5f84d01e7d39 100644
--- a/drivers/cxl/acpi.c
+++ b/drivers/cxl/acpi.c
@@ -61,7 +61,7 @@ u64 cxl_do_xormap_calc(struct cxl_cxims_data *cximsd, u64 addr, int hbiw)
* bits results in val==0, if odd the XOR result is val==1.
*/
- for (int i = 0; i < cximsd->nr_maps; i++) {
+ for (int i = 0; i < nr_maps_to_apply; i++) {
if (!cximsd->xormaps[i])
continue;
pos = __ffs(cximsd->xormaps[i]);
--
2.34.1
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH 1/1] cxl/acpi: Fix using extra xormaps for address translation
2026-01-15 13:16 [PATCH 1/1] cxl/acpi: Fix using extra xormaps for address translation Li Ming
@ 2026-01-15 16:12 ` Dave Jiang
2026-01-15 17:56 ` Alison Schofield
1 sibling, 0 replies; 4+ messages in thread
From: Dave Jiang @ 2026-01-15 16:12 UTC (permalink / raw)
To: Li Ming, dave, jonathan.cameron, alison.schofield,
vishal.l.verma, ira.weiny, dan.j.williams
Cc: linux-cxl, linux-kernel
On 1/15/26 6:16 AM, Li Ming wrote:
> CXL r3.1 Table 9-22 defines a fixed number of XOR maps for interleave
> arithmetic based on the number of interleave ways, as reflected by
> hbiw_to_nr_maps[].
>
> The current implementation applies all XOR maps during address
> translation, but it should use the spec-defined number of XOR maps
> derived from the number of interleave ways.
Can you add what is the current impact without applying the fix pls?
DJ
>
> Fixes: 3b2fedcd75e3 ("cxl: Restore XOR'd position bits during address translation")
> Signed-off-by: Li Ming <ming.li@zohomail.com>
> ---
> drivers/cxl/acpi.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/cxl/acpi.c b/drivers/cxl/acpi.c
> index 77ac940e3013..5f84d01e7d39 100644
> --- a/drivers/cxl/acpi.c
> +++ b/drivers/cxl/acpi.c
> @@ -61,7 +61,7 @@ u64 cxl_do_xormap_calc(struct cxl_cxims_data *cximsd, u64 addr, int hbiw)
> * bits results in val==0, if odd the XOR result is val==1.
> */
>
> - for (int i = 0; i < cximsd->nr_maps; i++) {
> + for (int i = 0; i < nr_maps_to_apply; i++) {
> if (!cximsd->xormaps[i])
> continue;
> pos = __ffs(cximsd->xormaps[i]);
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH 1/1] cxl/acpi: Fix using extra xormaps for address translation
2026-01-15 13:16 [PATCH 1/1] cxl/acpi: Fix using extra xormaps for address translation Li Ming
2026-01-15 16:12 ` Dave Jiang
@ 2026-01-15 17:56 ` Alison Schofield
2026-01-16 9:54 ` Li Ming
1 sibling, 1 reply; 4+ messages in thread
From: Alison Schofield @ 2026-01-15 17:56 UTC (permalink / raw)
To: Li Ming
Cc: dave, jonathan.cameron, dave.jiang, vishal.l.verma, ira.weiny,
dan.j.williams, linux-cxl, linux-kernel
On Thu, Jan 15, 2026 at 09:16:48PM +0800, Li Ming wrote:
> CXL r3.1 Table 9-22 defines a fixed number of XOR maps for interleave
> arithmetic based on the number of interleave ways, as reflected by
> hbiw_to_nr_maps[].
>
> The current implementation applies all XOR maps during address
> translation, but it should use the spec-defined number of XOR maps
> derived from the number of interleave ways.
>
> Fixes: 3b2fedcd75e3 ("cxl: Restore XOR'd position bits during address translation")
> Signed-off-by: Li Ming <ming.li@zohomail.com>
Ming,
I previously sent similar patch and then recalled it for the reason
described in this thread.
https://lore.kernel.org/linux-cxl/26c7ca6f-eb5f-4f20-8dba-e62e43e3e1b5@intel.com
The above was all before
"4fe516d2ad1a6 cxl/acpi: Make the XOR calculations available for testing"
where I added input param checking and introduced nr_maps_to_apply.
I don't think it is necessary to use nr_maps_to_apply in the
for loop once the params have been validated. So maybe this patch
is a cleanup on commit 4fe516d2ad1a6 rather than a fix.
Please take a look again and let me know if I'm wrong again ;)
--Alison
> ---
> drivers/cxl/acpi.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/cxl/acpi.c b/drivers/cxl/acpi.c
> index 77ac940e3013..5f84d01e7d39 100644
> --- a/drivers/cxl/acpi.c
> +++ b/drivers/cxl/acpi.c
> @@ -61,7 +61,7 @@ u64 cxl_do_xormap_calc(struct cxl_cxims_data *cximsd, u64 addr, int hbiw)
> * bits results in val==0, if odd the XOR result is val==1.
> */
>
> - for (int i = 0; i < cximsd->nr_maps; i++) {
> + for (int i = 0; i < nr_maps_to_apply; i++) {
> if (!cximsd->xormaps[i])
> continue;
> pos = __ffs(cximsd->xormaps[i]);
> --
> 2.34.1
>
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH 1/1] cxl/acpi: Fix using extra xormaps for address translation
2026-01-15 17:56 ` Alison Schofield
@ 2026-01-16 9:54 ` Li Ming
0 siblings, 0 replies; 4+ messages in thread
From: Li Ming @ 2026-01-16 9:54 UTC (permalink / raw)
To: Alison Schofield, dave.jiang
Cc: dave, jonathan.cameron, vishal.l.verma, ira.weiny,
dan.j.williams, linux-cxl, linux-kernel
在 2026/1/16 01:56, Alison Schofield 写道:
> On Thu, Jan 15, 2026 at 09:16:48PM +0800, Li Ming wrote:
>> CXL r3.1 Table 9-22 defines a fixed number of XOR maps for interleave
>> arithmetic based on the number of interleave ways, as reflected by
>> hbiw_to_nr_maps[].
>>
>> The current implementation applies all XOR maps during address
>> translation, but it should use the spec-defined number of XOR maps
>> derived from the number of interleave ways.
>>
>> Fixes: 3b2fedcd75e3 ("cxl: Restore XOR'd position bits during address translation")
>> Signed-off-by: Li Ming <ming.li@zohomail.com>
> Ming,
>
> I previously sent similar patch and then recalled it for the reason
> described in this thread.
> https://lore.kernel.org/linux-cxl/26c7ca6f-eb5f-4f20-8dba-e62e43e3e1b5@intel.com
>
> The above was all before
> "4fe516d2ad1a6 cxl/acpi: Make the XOR calculations available for testing"
> where I added input param checking and introduced nr_maps_to_apply.
>
> I don't think it is necessary to use nr_maps_to_apply in the
> for loop once the params have been validated. So maybe this patch
> is a cleanup on commit 4fe516d2ad1a6 rather than a fix.
>
> Please take a look again and let me know if I'm wrong again ;)
>
> --Alison
Hi Alison,
Thanks for pointing it out, you are right, it is not a bug. After
reviewing cxl_parse_cxims(), the cxmisd->xormaps stores the correct
number of xor maps based on the interleave ways. In fact,
nr_maps_to_apply is equal to cximsd->nr_maps in cxl_do_xormap_calc(), So
it is not needed to change it.
Hi Alison and Dave,
Please ignore this patch, apologies for the noise.
Thanks
Ming
>
>> ---
>> drivers/cxl/acpi.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/cxl/acpi.c b/drivers/cxl/acpi.c
>> index 77ac940e3013..5f84d01e7d39 100644
>> --- a/drivers/cxl/acpi.c
>> +++ b/drivers/cxl/acpi.c
>> @@ -61,7 +61,7 @@ u64 cxl_do_xormap_calc(struct cxl_cxims_data *cximsd, u64 addr, int hbiw)
>> * bits results in val==0, if odd the XOR result is val==1.
>> */
>>
>> - for (int i = 0; i < cximsd->nr_maps; i++) {
>> + for (int i = 0; i < nr_maps_to_apply; i++) {
>> if (!cximsd->xormaps[i])
>> continue;
>> pos = __ffs(cximsd->xormaps[i]);
>> --
>> 2.34.1
>>
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2026-01-16 9:54 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-01-15 13:16 [PATCH 1/1] cxl/acpi: Fix using extra xormaps for address translation Li Ming
2026-01-15 16:12 ` Dave Jiang
2026-01-15 17:56 ` Alison Schofield
2026-01-16 9:54 ` Li Ming
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®