[PATCH] mm/khugepaged: return -EAGAIN for SCAN_PAGE_HAS_PRIVATE in MADV_COLLAPSE

[PATCH] mm/khugepaged: return -EAGAIN for SCAN_PAGE_HAS_PRIVATE in MADV_COLLAPSE

[Vineet Agarwal]

MADV_COLLAPSE uses errno values to provide actionable feedback to
userspace. Temporary resource constraints are mapped to -EAGAIN so the
caller may retry, while intrinsic failures of the specified range are
mapped to -EINVAL.

collapse_file() returns SCAN_PAGE_HAS_PRIVATE when
filemap_release_folio() fails while isolating file-backed folios for
collapse. This currently falls through the default case in
madvise_collapse_errno() and is reported to userspace as -EINVAL.

However, filemap_release_folio() failure commonly reflects temporary
folio state rather than a permanently uncollapsible range.

For example, ext4 returns false when a folio still has dirty
journalled data, btrfs returns false for dirty or writeback folios
before extent state release, and NFS may return false while reclaiming
filesystem-private folio state.

In such cases, retrying MADV_COLLAPSE after writeback, reclaim or
journal progress may succeed. This matches the existing -EAGAIN
handling for SCAN_PAGE_DIRTY_OR_WRITEBACK and other transient collapse
failures more closely than -EINVAL.

Therefore, map SCAN_PAGE_HAS_PRIVATE to -EAGAIN so userspace receives
retryable feedback for this temporary failure path.

Signed-off-by: Vineet Agarwal <agarwal.vineet2006@gmail.com>
---
 mm/khugepaged.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index b8452dbdb043..58f55115d7d8 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -2808,6 +2808,7 @@ static int madvise_collapse_errno(enum scan_result r)
 	case SCAN_PAGE_LRU:
 	case SCAN_DEL_PAGE_LRU:
 	case SCAN_PAGE_FILLED:
+	case SCAN_PAGE_HAS_PRIVATE:
 	case SCAN_PAGE_DIRTY_OR_WRITEBACK:
 		return -EAGAIN;
 	/*
-- 
2.54.0

Re: [PATCH] mm/khugepaged: return -EAGAIN for SCAN_PAGE_HAS_PRIVATE in MADV_COLLAPSE

[Dev Jain]

On 29/04/26 7:34 pm, Vineet Agarwal wrote:
> MADV_COLLAPSE uses errno values to provide actionable feedback to
> userspace. Temporary resource constraints are mapped to -EAGAIN so the
> caller may retry, while intrinsic failures of the specified range are
> mapped to -EINVAL.
> 
> collapse_file() returns SCAN_PAGE_HAS_PRIVATE when
> filemap_release_folio() fails while isolating file-backed folios for
> collapse. This currently falls through the default case in
> madvise_collapse_errno() and is reported to userspace as -EINVAL.
> 
> However, filemap_release_folio() failure commonly reflects temporary
> folio state rather than a permanently uncollapsible range.
> 
> For example, ext4 returns false when a folio still has dirty
> journalled data, btrfs returns false for dirty or writeback folios
> before extent state release, and NFS may return false while reclaiming
> filesystem-private folio state.
> 
> In such cases, retrying MADV_COLLAPSE after writeback, reclaim or
> journal progress may succeed. This matches the existing -EAGAIN
> handling for SCAN_PAGE_DIRTY_OR_WRITEBACK and other transient collapse
> failures more closely than -EINVAL.
> 
> Therefore, map SCAN_PAGE_HAS_PRIVATE to -EAGAIN so userspace receives
> retryable feedback for this temporary failure path.
> 
> Signed-off-by: Vineet Agarwal <agarwal.vineet2006@gmail.com>
> ---
>  mm/khugepaged.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> index b8452dbdb043..58f55115d7d8 100644
> --- a/mm/khugepaged.c
> +++ b/mm/khugepaged.c
> @@ -2808,6 +2808,7 @@ static int madvise_collapse_errno(enum scan_result r)
>  	case SCAN_PAGE_LRU:
>  	case SCAN_DEL_PAGE_LRU:
>  	case SCAN_PAGE_FILLED:
> +	case SCAN_PAGE_HAS_PRIVATE:
>  	case SCAN_PAGE_DIRTY_OR_WRITEBACK:
>  		return -EAGAIN;
>  	/*

https://lore.kernel.org/all/4ab54ae0-2607-443d-8698-788d8e951bdd@arm.com/

I had mentioned at that time, other error codes may fit the -EAGAIN
(or something else) definition as well, for example, probably SCAN_TRUNCATED.

Could you please audit other error codes as well?

Anyhow to your patch here:

Reviewed-by: Dev Jain <dev.jain@arm.com>

Re: [PATCH] mm/khugepaged: return -EAGAIN for SCAN_PAGE_HAS_PRIVATE in MADV_COLLAPSE

[David Hildenbrand (Arm)]

On 4/29/26 16:20, Dev Jain wrote:
> 
> 
> On 29/04/26 7:34 pm, Vineet Agarwal wrote:
>> MADV_COLLAPSE uses errno values to provide actionable feedback to
>> userspace. Temporary resource constraints are mapped to -EAGAIN so the
>> caller may retry, while intrinsic failures of the specified range are
>> mapped to -EINVAL.
>>
>> collapse_file() returns SCAN_PAGE_HAS_PRIVATE when
>> filemap_release_folio() fails while isolating file-backed folios for
>> collapse. This currently falls through the default case in
>> madvise_collapse_errno() and is reported to userspace as -EINVAL.
>>
>> However, filemap_release_folio() failure commonly reflects temporary
>> folio state rather than a permanently uncollapsible range.
>>
>> For example, ext4 returns false when a folio still has dirty
>> journalled data, btrfs returns false for dirty or writeback folios
>> before extent state release, and NFS may return false while reclaiming
>> filesystem-private folio state.
>>
>> In such cases, retrying MADV_COLLAPSE after writeback, reclaim or
>> journal progress may succeed. This matches the existing -EAGAIN
>> handling for SCAN_PAGE_DIRTY_OR_WRITEBACK and other transient collapse
>> failures more closely than -EINVAL.
>>
>> Therefore, map SCAN_PAGE_HAS_PRIVATE to -EAGAIN so userspace receives
>> retryable feedback for this temporary failure path.
>>
>> Signed-off-by: Vineet Agarwal <agarwal.vineet2006@gmail.com>
>> ---
>>  mm/khugepaged.c | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
>> index b8452dbdb043..58f55115d7d8 100644
>> --- a/mm/khugepaged.c
>> +++ b/mm/khugepaged.c
>> @@ -2808,6 +2808,7 @@ static int madvise_collapse_errno(enum scan_result r)
>>  	case SCAN_PAGE_LRU:
>>  	case SCAN_DEL_PAGE_LRU:
>>  	case SCAN_PAGE_FILLED:
>> +	case SCAN_PAGE_HAS_PRIVATE:
>>  	case SCAN_PAGE_DIRTY_OR_WRITEBACK:
>>  		return -EAGAIN;
>>  	/*
> 
> https://lore.kernel.org/all/4ab54ae0-2607-443d-8698-788d8e951bdd@arm.com/
> 
> I had mentioned at that time, other error codes may fit the -EAGAIN
> (or something else) definition as well, for example, probably SCAN_TRUNCATED.

In folio_split(), we such events as a mixture of EAGAIN and EBUSY.
filemap_release_folio() errors are, for example, handles as EBUSY.

So in that regard, using EAGAIN here (as we don't use EBUSY) makes sense for
SCAN_PAGE_HAS_PRIVATE.

-- 
Cheers,

David

Re: [PATCH] mm/khugepaged: return -EAGAIN for SCAN_PAGE_HAS_PRIVATE in MADV_COLLAPSE

[Andrew Morton]

On Wed, 29 Apr 2026 19:34:34 +0530 Vineet Agarwal <agarwal.vineet2006@gmail.com> wrote:

> MADV_COLLAPSE uses errno values to provide actionable feedback to
> userspace. Temporary resource constraints are mapped to -EAGAIN so the
> caller may retry, while intrinsic failures of the specified range are
> mapped to -EINVAL.
> 
> collapse_file() returns SCAN_PAGE_HAS_PRIVATE when
> filemap_release_folio() fails while isolating file-backed folios for
> collapse. This currently falls through the default case in
> madvise_collapse_errno() and is reported to userspace as -EINVAL.
> 
> However, filemap_release_folio() failure commonly reflects temporary
> folio state rather than a permanently uncollapsible range.
> 
> For example, ext4 returns false when a folio still has dirty
> journalled data, btrfs returns false for dirty or writeback folios
> before extent state release, and NFS may return false while reclaiming
> filesystem-private folio state.
> 
> In such cases, retrying MADV_COLLAPSE after writeback, reclaim or
> journal progress may succeed. This matches the existing -EAGAIN
> handling for SCAN_PAGE_DIRTY_OR_WRITEBACK and other transient collapse
> failures more closely than -EINVAL.
> 
> Therefore, map SCAN_PAGE_HAS_PRIVATE to -EAGAIN so userspace receives
> retryable feedback for this temporary failure path.

Seems very reasonable to me.

The madvise(2) manpage could be a lot more helpful here.

	EAGAIN A kernel resource was temporarily unavailable.

and that's it!