Re: [PATCH 12/23] bpf: handle the compat string in bpf_trace_copy_string better

From: Andrew Morton <akpm@linux-foundation.org>
To: Christoph Hellwig <hch@lst.de>
Cc: x86@kernel.org, Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Masami Hiramatsu <mhiramat@kernel.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	linux-parisc@vger.kernel.org, linux-um@lists.infradead.org,
	netdev@vger.kernel.org, bpf@vger.kernel.org, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH 12/23] bpf: handle the compat string in bpf_trace_copy_string better
Date: Wed, 27 May 2020 19:04:32 -0700	[thread overview]
Message-ID: <20200527190432.e4af1fba00c13cb1421f5a37@linux-foundation.org> (raw)
In-Reply-To: <20200521152301.2587579-13-hch@lst.de>

On Thu, 21 May 2020 17:22:50 +0200 Christoph Hellwig <hch@lst.de> wrote:

> User the proper helper for kernel or userspace addresses based on
> TASK_SIZE instead of the dangerous strncpy_from_unsafe function.
> 
> ...
>
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -331,8 +331,11 @@ static void bpf_trace_copy_string(char *buf, void *unsafe_ptr, char fmt_ptype,
>  	switch (fmt_ptype) {
>  	case 's':
>  #ifdef CONFIG_ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
> -		strncpy_from_unsafe(buf, unsafe_ptr, bufsz);
> -		break;
> +		if ((unsigned long)unsafe_ptr < TASK_SIZE) {
> +			strncpy_from_user_nofault(buf, user_ptr, bufsz);
> +			break;
> +		}
> +		fallthrough;
>  #endif
>  	case 'k':
>  		strncpy_from_kernel_nofault(buf, unsafe_ptr, bufsz);

Another user of strncpy_from_unsafe() has popped up in linux-next's
bpf.  I did the below, but didn't try very hard - it's probably wrong
if CONFIG_ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE=n?

Anyway, please take a look at all the bpf_trace.c changes in
linux-next.


From: Andrew Morton <akpm@linux-foundation.org>
Subject: bpf:bpf_seq_printf(): handle potentially unsafe format string better

User the proper helper for kernel or userspace addresses based on
TASK_SIZE instead of the dangerous strncpy_from_unsafe function.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 kernel/trace/bpf_trace.c |   13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

--- a/kernel/trace/bpf_trace.c~xxx
+++ a/kernel/trace/bpf_trace.c
@@ -588,15 +588,22 @@ BPF_CALL_5(bpf_seq_printf, struct seq_fi
 		}
 
 		if (fmt[i] == 's') {
+			void *unsafe_ptr;
+
 			/* try our best to copy */
 			if (memcpy_cnt >= MAX_SEQ_PRINTF_MAX_MEMCPY) {
 				err = -E2BIG;
 				goto out;
 			}
 
-			err = strncpy_from_unsafe(bufs->buf[memcpy_cnt],
-						  (void *) (long) args[fmt_cnt],
-						  MAX_SEQ_PRINTF_STR_LEN);
+			unsafe_ptr = (void *)(long)args[fmt_cnt];
+			if ((unsigned long)unsafe_ptr < TASK_SIZE) {
+				err = strncpy_from_user_nofault(
+					bufs->buf[memcpy_cnt], unsafe_ptr,
+					MAX_SEQ_PRINTF_STR_LEN);
+			} else {
+				err = -EFAULT;
+			}
 			if (err < 0)
 				bufs->buf[memcpy_cnt][0] = '\0';
 			params[fmt_cnt] = (u64)(long)bufs->buf[memcpy_cnt];
_

