From: Leon Romanovsky <leon@kernel.org>
To: Sung Byeongchan <tjdqudcks0424@naver.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>, linux-rdma@vger.kernel.org
Subject: Re: [PATCH] RDMA/iwcm: restrict MAPINFO to administrators
Date: Mon, 9 Nov 2026 10:23:49 +0200 [thread overview]
Message-ID: <20261109082349.GB8635@unreal> (raw)
In-Reply-To: <20261107093338.618394-1-tjdqudcks0424@naver.com>
On Sat, Nov 07, 2026 at 06:33:38PM +0900, Sung Byeongchan wrote:
>
> RDMA_NL_IWPM_MAPINFO announces the userspace iWARP port mapper and asks
> the kernel to send its live mapping table. Its callback registration does
> not request an administrative permission check, so an ordinary local
> NETLINK_RDMA sender can use the public mapper name and ABI version to
> replace the global daemon portid.
>
> With a production SIW mapping present, an unprivileged claimant received
> the complete local and mapped sockaddr records in three of three fresh
> KASAN boots. A second independent unprivileged socket could repeat the
> claim. Wrong-name and old-version controls were rejected.
>
> Set RDMA_NL_ADMIN_PERM on only the MAPINFO callback. The common RDMA
> netlink receiver then requires CAP_NET_ADMIN before the callback changes
> the daemon identity or serializes the mapping table.
>
> The fixed kernel rejected the valid unprivileged request with -EPERM and
> zero records in three of three matched boots. Legitimate root mapper
> registration, live SIW mapping, daemon recovery, mapping removal and
> post-remove cleanup all passed their baseline and fixed controls.
>
> The demonstrated impact is disclosure of local iWARP endpoint metadata
> and displacement of the trusted mapper identity. Uninitialized memory,
> controlled read or write, RCE and LPE were not demonstrated.
1. People still read commit messages, so please read yours before submitting.
2. Write a proper commit message; don't submit AI-generated slop.
3. Include the full kernel stack trace from dmesg in the commit message.
4. Add RDMA_NL_ADMIN_PERM to all IWPM commands.
Thanks
>
> Fixes: b493d91d333e ("iwcm: common code for port mapper")
> Assisted-by: OpenAI Codex
>
> Signed-off-by: Sung Byeongchan <tjdqudcks0424@naver.com>
> ---
> drivers/infiniband/core/iwcm.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
> diff --git a/drivers/infiniband/core/iwcm.c b/drivers/infiniband/core/iwcm.c
> index 4b05867dad895..8a3b5aa2dc5f7 100644
> --- a/drivers/infiniband/core/iwcm.c
> +++ b/drivers/infiniband/core/iwcm.c
> @@ -86,7 +86,10 @@ static struct rdma_nl_cbs iwcm_nl_cb_table[RDMA_NL_IWPM_NUM_OPS] = {
> [RDMA_NL_IWPM_QUERY_MAPPING] = {.dump = iwpm_add_and_query_mapping_cb},
> [RDMA_NL_IWPM_REMOTE_INFO] = {.dump = iwpm_remote_info_cb},
> [RDMA_NL_IWPM_HANDLE_ERR] = {.dump = iwpm_mapping_error_cb},
> - [RDMA_NL_IWPM_MAPINFO] = {.dump = iwpm_mapping_info_cb},
> + [RDMA_NL_IWPM_MAPINFO] = {
> + .dump = iwpm_mapping_info_cb,
> + .flags = RDMA_NL_ADMIN_PERM,
> + },
> [RDMA_NL_IWPM_MAPINFO_NUM] = {.dump = iwpm_ack_mapping_info_cb},
> [RDMA_NL_IWPM_HELLO] = {.dump = iwpm_hello_cb}
> };
>
> base-commit: d48723711bae9b8cdc56719c448ba42fef28cbee
> --
> 2.43.0
prev parent reply other threads:[~2026-11-09 8:23 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-07 9:33 [PATCH] RDMA/iwcm: restrict MAPINFO to administrators Sung Byeongchan
2026-11-07 9:45 ` sashiko-bot
2026-11-09 8:23 ` Leon Romanovsky [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20261109082349.GB8635@unreal \
--to=leon@kernel.org \
--cc=jgg@ziepe.ca \
--cc=linux-rdma@vger.kernel.org \
--cc=tjdqudcks0424@naver.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox