[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAB9dFdvLNPTkgzv5GJ9Mwdtpz-sk0XYWQ25hqxcpT_NPufnuKQ@mail.gmail.com>
Date: Wed, 30 Mar 2022 11:15:26 -0300
From: Marc Dionne <marc.dionne@...istor.com>
To: Xiaolong Huang <butterflyhuangxx@...il.com>
Cc: David Howells <dhowells@...hat.com>,
David Miller <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>, pabeni@...hat.com,
linux-afs@...ts.infradead.org, netdev <netdev@...r.kernel.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] rxrpc: fix some null-ptr-deref bugs in server_key.c
On Tue, Mar 22, 2022 at 1:28 PM Xiaolong Huang
<butterflyhuangxx@...il.com> wrote:
>
> Some function calls are not implemented in rxrpc_no_security, there are
> preparse_server_key, free_preparse_server_key and destroy_server_key.
> When rxrpc security type is rxrpc_no_security, user can easily trigger a
> null-ptr-deref bug via ioctl. So judgment should be added to prevent it
>
> The crash log:
> user@...kaller:~$ ./rxrpc_preparse_s
> [ 37.956878][T15626] BUG: kernel NULL pointer dereference, address: 0000000000000000
> [ 37.957645][T15626] #PF: supervisor instruction fetch in kernel mode
> [ 37.958229][T15626] #PF: error_code(0x0010) - not-present page
> [ 37.958762][T15626] PGD 4aadf067 P4D 4aadf067 PUD 4aade067 PMD 0
> [ 37.959321][T15626] Oops: 0010 [#1] PREEMPT SMP
> [ 37.959739][T15626] CPU: 0 PID: 15626 Comm: rxrpc_preparse_ Not tainted 5.17.0-01442-gb47d5a4f6b8d #43
> [ 37.960588][T15626] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-1ubuntu1 04/01/2014
> [ 37.961474][T15626] RIP: 0010:0x0
> [ 37.961787][T15626] Code: Unable to access opcode bytes at RIP 0xffffffffffffffd6.
> [ 37.962480][T15626] RSP: 0018:ffffc9000d9abdc0 EFLAGS: 00010286
> [ 37.963018][T15626] RAX: ffffffff84335200 RBX: ffff888012a1ce80 RCX: 0000000000000000
> [ 37.963727][T15626] RDX: 0000000000000000 RSI: ffffffff84a736dc RDI: ffffc9000d9abe48
> [ 37.964425][T15626] RBP: ffffc9000d9abe48 R08: 0000000000000000 R09: 0000000000000002
> [ 37.965118][T15626] R10: 000000000000000a R11: f000000000000000 R12: ffff888013145680
> [ 37.965836][T15626] R13: 0000000000000000 R14: ffffffffffffffec R15: ffff8880432aba80
> [ 37.966441][T15626] FS: 00007f2177907700(0000) GS:ffff88803ec00000(0000) knlGS:0000000000000000
> [ 37.966979][T15626] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [ 37.967384][T15626] CR2: ffffffffffffffd6 CR3: 000000004aaf1000 CR4: 00000000000006f0
> [ 37.967864][T15626] Call Trace:
> [ 37.968062][T15626] <TASK>
> [ 37.968240][T15626] rxrpc_preparse_s+0x59/0x90
> [ 37.968541][T15626] key_create_or_update+0x174/0x510
> [ 37.968863][T15626] __x64_sys_add_key+0x139/0x1d0
> [ 37.969165][T15626] do_syscall_64+0x35/0xb0
> [ 37.969451][T15626] entry_SYSCALL_64_after_hwframe+0x44/0xae
> [ 37.969824][T15626] RIP: 0033:0x43a1f9
>
> Signed-off-by: Xiaolong Huang <butterflyhuangxx@...il.com>
> Tested-by: Xiaolong Huang <butterflyhuangxx@...il.com>
> ---
> net/rxrpc/server_key.c | 7 +++++--
> 1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/net/rxrpc/server_key.c b/net/rxrpc/server_key.c
> index ead3471307ee..ee269e0e6ee8 100644
> --- a/net/rxrpc/server_key.c
> +++ b/net/rxrpc/server_key.c
> @@ -84,6 +84,9 @@ static int rxrpc_preparse_s(struct key_preparsed_payload *prep)
>
> prep->payload.data[1] = (struct rxrpc_security *)sec;
>
> + if (!sec->preparse_server_key)
> + return -EINVAL;
> +
> return sec->preparse_server_key(prep);
> }
>
> @@ -91,7 +94,7 @@ static void rxrpc_free_preparse_s(struct key_preparsed_payload *prep)
> {
> const struct rxrpc_security *sec = prep->payload.data[1];
>
> - if (sec)
> + if (sec && sec->free_preparse_server_key)
> sec->free_preparse_server_key(prep);
> }
>
> @@ -99,7 +102,7 @@ static void rxrpc_destroy_s(struct key *key)
> {
> const struct rxrpc_security *sec = key->payload.data[1];
>
> - if (sec)
> + if (sec && sec->destroy_server_key)
> sec->destroy_server_key(key);
> }
>
>
> base-commit: b47d5a4f6b8d42f8a8fbe891b36215e4fddc53be
> --
> 2.25.1
Actually ran into this while experimenting; looks fine and fixes the problem.
Acked-by: Marc Dionne <marc.dionne@...istor.com>
Marc
Powered by blists - more mailing lists