[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-id: <164806948116.6096.12331736937963571485@noble.neil.brown.name>
Date: Thu, 24 Mar 2022 08:04:41 +1100
From: "NeilBrown" <neilb@...e.de>
To: xkernel.wang@...mail.com
Cc: trond.myklebust@...merspace.com, anna.schumaker@...app.com,
linux-nfs@...r.kernel.org, linux-kernel@...r.kernel.org,
"Xiaoke Wang" <xkernel.wang@...mail.com>
Subject: Re: [PATCH] NFS: check the return value of mempool_alloc()
On Wed, 23 Mar 2022, xkernel.wang@...mail.com wrote:
> From: Xiaoke Wang <xkernel.wang@...mail.com>
>
> The check was first removed in 518662e ("NFS: fix usage of mempools.")
> as the passed GFP flags is `GFP_NOIO`.
> While now the flag is changed to `GFP_KERNEL` by 2b17d72 ("NFS: Clean
> up writeback code"), so it is better to check it.
no. GFP_KERNEL is not that different from GFP_NOIO.
mempool_alloc() can only fail with __GFP_DIRECT_RECLAIM is not passed.
Please try to understand the code before you change it.
NeilBrown
>
> Signed-off-by: Xiaoke Wang <xkernel.wang@...mail.com>
> ---
> fs/nfs/write.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/fs/nfs/write.c b/fs/nfs/write.c
> index eae9bf1..831fad9 100644
> --- a/fs/nfs/write.c
> +++ b/fs/nfs/write.c
> @@ -106,8 +106,10 @@ static struct nfs_pgio_header *nfs_writehdr_alloc(void)
> {
> struct nfs_pgio_header *p = mempool_alloc(nfs_wdata_mempool, GFP_KERNEL);
>
> - memset(p, 0, sizeof(*p));
> - p->rw_mode = FMODE_WRITE;
> + if (p) {
> + memset(p, 0, sizeof(*p));
> + p->rw_mode = FMODE_WRITE;
> + }
> return p;
> }
>
> --
>
>
Powered by blists - more mailing lists