2023-12-28 06:18:59

by Barry Song

Subject: [PATCH] mm: zsmalloc: return -ENOSPC rather than -EINVAL in zs_malloc while size is too large

This is the case the "compressed" data is larger than the original data,
it is better to return -ENOSPC which can help zswap record a poor compr
rather than an invalid request. Then we get more friendly counting for
reject_compress_poor in debugfs.

bool zswap_store(struct folio *folio)
{
...
ret = zpool_malloc(zpool, dlen, gfp, &handle);
if (ret == -ENOSPC) {
zswap_reject_compress_poor++;
goto put_dstmem;
}
if (ret) {
zswap_reject_alloc_fail++;
goto put_dstmem;
}
...
}

Also, zbud_alloc() and z3fold_alloc() are returning ENOSPC in the same
case, eg
static int z3fold_alloc(struct z3fold_pool *pool, size_t size, gfp_t gfp,
unsigned long *handle)
{
...
if (!size || (gfp & __GFP_HIGHMEM))
return -EINVAL;

if (size > PAGE_SIZE)
return -ENOSPC;
...
}

Signed-off-by: Barry Song <[email protected]>
---
mm/zsmalloc.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c
index b1c0dad7f4cf..c937635e0ad1 100644
--- a/mm/zsmalloc.c
+++ b/mm/zsmalloc.c
@@ -1364,9 +1364,12 @@ unsigned long zs_malloc(struct zs_pool *pool, size_t size, gfp_t gfp)
int newfg;
struct zspage *zspage;

- if (unlikely(!size || size > ZS_MAX_ALLOC_SIZE))
+ if (unlikely(!size))
return (unsigned long)ERR_PTR(-EINVAL);

+ if (unlikely(size > ZS_MAX_ALLOC_SIZE))
+ return (unsigned long)ERR_PTR(-ENOSPC);
+
handle = cache_alloc_handle(pool, gfp);
if (!handle)
return (unsigned long)ERR_PTR(-ENOMEM);
--
2.34.1

2023-12-28 06:23:55

by Chengming Zhou

Subject: Re: [PATCH] mm: zsmalloc: return -ENOSPC rather than -EINVAL in zs_malloc while size is too large

On 2023/12/28 14:18, Barry Song wrote:
> This is the case the "compressed" data is larger than the original data,
> it is better to return -ENOSPC which can help zswap record a poor compr
> rather than an invalid request. Then we get more friendly counting for
> reject_compress_poor in debugfs.
>
> bool zswap_store(struct folio *folio)
> {
> ...
> ret = zpool_malloc(zpool, dlen, gfp, &handle);
> if (ret == -ENOSPC) {
> zswap_reject_compress_poor++;
> goto put_dstmem;
> }
> if (ret) {
> zswap_reject_alloc_fail++;
> goto put_dstmem;
> }
> ...
> }
>
> Also, zbud_alloc() and z3fold_alloc() are returning ENOSPC in the same
> case, eg
> static int z3fold_alloc(struct z3fold_pool *pool, size_t size, gfp_t gfp,
> unsigned long *handle)
> {
> ...
> if (!size || (gfp & __GFP_HIGHMEM))
> return -EINVAL;
>
> if (size > PAGE_SIZE)
> return -ENOSPC;
> ...
> }
>
> Signed-off-by: Barry Song <[email protected]>

Looks good to me.

Reviewed-by: Chengming Zhou <[email protected]>

Thanks.

> ---
> mm/zsmalloc.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c
> index b1c0dad7f4cf..c937635e0ad1 100644
> --- a/mm/zsmalloc.c
> +++ b/mm/zsmalloc.c
> @@ -1364,9 +1364,12 @@ unsigned long zs_malloc(struct zs_pool *pool, size_t size, gfp_t gfp)
> int newfg;
> struct zspage *zspage;
>
> - if (unlikely(!size || size > ZS_MAX_ALLOC_SIZE))
> + if (unlikely(!size))
> return (unsigned long)ERR_PTR(-EINVAL);
>
> + if (unlikely(size > ZS_MAX_ALLOC_SIZE))
> + return (unsigned long)ERR_PTR(-ENOSPC);
> +
> handle = cache_alloc_handle(pool, gfp);
> if (!handle)
> return (unsigned long)ERR_PTR(-ENOMEM);

2023-12-28 19:19:28

by Nhat Pham

Subject: Re: [PATCH] mm: zsmalloc: return -ENOSPC rather than -EINVAL in zs_malloc while size is too large

On Wed, Dec 27, 2023 at 10:18 PM Barry Song <[email protected]> wrote:
>
> This is the case the "compressed" data is larger than the original data,
> it is better to return -ENOSPC which can help zswap record a poor compr
> rather than an invalid request. Then we get more friendly counting for
> reject_compress_poor in debugfs.
>
> bool zswap_store(struct folio *folio)
> {
> ...
> ret = zpool_malloc(zpool, dlen, gfp, &handle);
> if (ret == -ENOSPC) {
> zswap_reject_compress_poor++;
> goto put_dstmem;
> }
> if (ret) {
> zswap_reject_alloc_fail++;
> goto put_dstmem;
> }
> ...
> }
>
> Also, zbud_alloc() and z3fold_alloc() are returning ENOSPC in the same
> case, eg
> static int z3fold_alloc(struct z3fold_pool *pool, size_t size, gfp_t gfp,
> unsigned long *handle)
> {
> ...
> if (!size || (gfp & __GFP_HIGHMEM))
> return -EINVAL;
>
> if (size > PAGE_SIZE)
> return -ENOSPC;
> ...
> }
>
> Signed-off-by: Barry Song <[email protected]>
> ---
> mm/zsmalloc.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c
> index b1c0dad7f4cf..c937635e0ad1 100644
> --- a/mm/zsmalloc.c
> +++ b/mm/zsmalloc.c
> @@ -1364,9 +1364,12 @@ unsigned long zs_malloc(struct zs_pool *pool, size_t size, gfp_t gfp)
> int newfg;
> struct zspage *zspage;
>
> - if (unlikely(!size || size > ZS_MAX_ALLOC_SIZE))
> + if (unlikely(!size))
> return (unsigned long)ERR_PTR(-EINVAL);
>
> + if (unlikely(size > ZS_MAX_ALLOC_SIZE))
> + return (unsigned long)ERR_PTR(-ENOSPC);
> +
> handle = cache_alloc_handle(pool, gfp);
> if (!handle)
> return (unsigned long)ERR_PTR(-ENOMEM);
> --
> 2.34.1
>

LGTM.
Reviewed-by: Nhat Pham <[email protected]>