Re: [PATCH RFC 01/11] mm: introduce hw_pte_t for PTE table storage

On 27/07/2026 5:58 pm, [email protected] wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [Medium] The header `<linux/pgtable_types.h>` incorrectly assumes `<asm/page.h>` universally defines `pte_t` and architecture opt-ins, leading to missing types on x86 and a structural inclusion hazard for opting in.
> --
>
> --- Patch [1]: [PATCH RFC 01/11] mm: introduce hw_pte_t for PTE table storage ---
> commit de503ea03294c1c9976ce045b5e820dbc0f34eba
> Author: Muhammad Usama Anjum <[email protected]>
>
> mm: introduce hw_pte_t for PTE table storage
>
> This commit introduces hw_pte_t as a generic name for a PTE table element.
> It defines it as an alias of pte_t by default, allowing architectures to
> retain their existing representation until they opt in to a distinct type.
>
>> diff --git a/include/linux/pgtable_types.h b/include/linux/pgtable_types.h
>> new file mode 100644
>> index 0000000000000..ed5dba42a4f8c
>> --- /dev/null
>> +++ b/include/linux/pgtable_types.h
>> @@ -0,0 +1,11 @@
>> +/* SPDX-License-Identifier: GPL-2.0 */
>> +#ifndef _LINUX_PGTABLE_TYPES_H
>> +#define _LINUX_PGTABLE_TYPES_H
>> +
>> +#include <asm/page.h>
>> +
>> +#ifndef __HAVE_ARCH_HW_PTE_T
>> +#define hw_pte_t pte_t
>> +#endif
>
> [Severity: Medium]
> Is <asm/page.h> guaranteed to define pte_t on all architectures before it is
> used here?
>
> On x86, pte_t is defined in <asm/pgtable_types.h>, which is not included by
> <asm/page.h>. This means pte_t might not be defined when the default hw_pte_t
> alias is created.
No. The commit message was imprecise here. hw_pte_t is defined as a macro alias.
pte_t does not need to be defined when this macro is defined; it only needs to
be defined where hw_pte_t is expanded. The current x86 consumers make pte_t
visible through <asm/pgtable_types.h> before expanding hw_pte_t.

>
> Could this also create an inclusion hazard for architectures opting in?
>
> If x86 were to opt into hw_pte_t by defining __HAVE_ARCH_HW_PTE_T in
> <asm/pgtable_types.h>, any files including <linux/pgtable_types.h> first would
> incorrectly alias hw_pte_t to pte_t. When <asm/pgtable_types.h> is
> subsequently included, its explicit definition of hw_pte_t would macro-expand
> to a conflicting redefinition of pte_t, breaking the build.
An architecture opting in must make both __HAVE_ARCH_HW_PTE_T and hw_pte_t visible
before the generic fallback is processed. Arm64 does this through asm/page.h which
includes asm/pgtable-types.h. A future x86 conversion would have to establish the
same ordering.

--
Thanks,
Usama

IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.