2023-11-22 23:40:29

by Kent Overstreet

Subject: [PATCH] rhashtable: Better error message on allocation failure

Memory allocation failures print backtraces by default, but when we're
running out of a rhashtable worker the backtrace is useless - it doesn't
tell us which hashtable the allocation failure was for.

This adds a dedicated warning that prints out functions from the
rhashtable params, which will be a bit more useful.

Signed-off-by: Kent Overstreet <[email protected]>
Cc: Thomas Graf <[email protected]>
Cc: Herbert Xu <[email protected]>
---
lib/rhashtable.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/lib/rhashtable.c b/lib/rhashtable.c
index 6ae2ba8e06a2..d3fce9c8989a 100644
--- a/lib/rhashtable.c
+++ b/lib/rhashtable.c
@@ -360,9 +360,14 @@ static int rhashtable_rehash_alloc(struct rhashtable *ht,

ASSERT_RHT_MUTEX(ht);

- new_tbl = bucket_table_alloc(ht, size, GFP_KERNEL);
- if (new_tbl == NULL)
+ new_tbl = bucket_table_alloc(ht, size, GFP_KERNEL|__GFP_NOWARN);
+ if (new_tbl == NULL) {
+ WARN("rhashtable bucket table allocation failure for %ps",
+ (void *) ht->p.hashfn ?:
+ (void *) ht->p.obj_hashfn ?:
+ (void *) ht->p.obj_cmpfn);
return -ENOMEM;
+ }

err = rhashtable_rehash_attach(ht, old_tbl, new_tbl);
if (err)
--
2.42.0

2023-11-23 08:33:34

by Herbert Xu

Subject: Re: [PATCH] rhashtable: Better error message on allocation failure

On Wed, Nov 22, 2023 at 06:32:16PM -0500, Kent Overstreet wrote:
> Memory allocation failures print backtraces by default, but when we're
> running out of a rhashtable worker the backtrace is useless - it doesn't
> tell us which hashtable the allocation failure was for.
>
> This adds a dedicated warning that prints out functions from the
> rhashtable params, which will be a bit more useful.
>
> Signed-off-by: Kent Overstreet <[email protected]>
> Cc: Thomas Graf <[email protected]>
> Cc: Herbert Xu <[email protected]>
> ---
> lib/rhashtable.c | 9 +++++++--
> 1 file changed, 7 insertions(+), 2 deletions(-)

Acked-by: Herbert Xu <[email protected]>

However, you'll need to repost this via [email protected] for
it to be merged.

Thanks,
--
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

2023-11-23 23:59:07

by Kent Overstreet

Subject: Re: [PATCH] rhashtable: Better error message on allocation failure

On Thu, Nov 23, 2023 at 04:32:47PM +0800, Herbert Xu wrote:
> On Wed, Nov 22, 2023 at 06:32:16PM -0500, Kent Overstreet wrote:
> > Memory allocation failures print backtraces by default, but when we're
> > running out of a rhashtable worker the backtrace is useless - it doesn't
> > tell us which hashtable the allocation failure was for.
> >
> > This adds a dedicated warning that prints out functions from the
> > rhashtable params, which will be a bit more useful.
> >
> > Signed-off-by: Kent Overstreet <[email protected]>
> > Cc: Thomas Graf <[email protected]>
> > Cc: Herbert Xu <[email protected]>
> > ---
> > lib/rhashtable.c | 9 +++++++--
> > 1 file changed, 7 insertions(+), 2 deletions(-)
>
> Acked-by: Herbert Xu <[email protected]>
>
> However, you'll need to repost this via [email protected] for
> it to be merged.

Thanks!