* [PATCH] CVE-2016-3075: Stack overflow in _nss_dns_getnetbyname_r [BZ #19879]
@ 2016-03-29 11:01 Florian Weimer
2016-03-29 13:49 ` Carlos O'Donell
2016-03-29 21:41 ` Joseph Myers
0 siblings, 2 replies; 4+ messages in thread
From: Florian Weimer @ 2016-03-29 11:01 UTC (permalink / raw)
To: GNU C Library
[-- Attachment #1: Type: text/plain, Size: 190 bytes --]
This is a minor security issue in nss_dns, triggered by a very long name
passed to getnetbyname.
The defensive copy is not needed because the name may not alias the
output buffer.
Florian
[-- Attachment #2: 0001-CVE-2016-3075-Stack-overflow-in-_nss_dns_getnetbynam.patch --]
[-- Type: text/x-patch, Size: 939 bytes --]
2016-03-29 Florian Weimer <fweimer@redhat.com>
[BZ #19879]
CVE-2016-3075
* resolv/nss_dns/dns-network.c (_nss_dns_getnetbyname_r): Do not
copy name.
diff --git a/resolv/nss_dns/dns-network.c b/resolv/nss_dns/dns-network.c
index 2eb2f67..8f301a7 100644
--- a/resolv/nss_dns/dns-network.c
+++ b/resolv/nss_dns/dns-network.c
@@ -118,17 +118,14 @@ _nss_dns_getnetbyname_r (const char *name, struct netent *result,
} net_buffer;
querybuf *orig_net_buffer;
int anslen;
- char *qbuf;
enum nss_status status;
if (__res_maybe_init (&_res, 0) == -1)
return NSS_STATUS_UNAVAIL;
- qbuf = strdupa (name);
-
net_buffer.buf = orig_net_buffer = (querybuf *) alloca (1024);
- anslen = __libc_res_nsearch (&_res, qbuf, C_IN, T_PTR, net_buffer.buf->buf,
+ anslen = __libc_res_nsearch (&_res, name, C_IN, T_PTR, net_buffer.buf->buf,
1024, &net_buffer.ptr, NULL, NULL, NULL, NULL);
if (anslen < 0)
{
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] CVE-2016-3075: Stack overflow in _nss_dns_getnetbyname_r [BZ #19879]
2016-03-29 11:01 [PATCH] CVE-2016-3075: Stack overflow in _nss_dns_getnetbyname_r [BZ #19879] Florian Weimer
@ 2016-03-29 13:49 ` Carlos O'Donell
2016-03-29 21:41 ` Joseph Myers
1 sibling, 0 replies; 4+ messages in thread
From: Carlos O'Donell @ 2016-03-29 13:49 UTC (permalink / raw)
To: Florian Weimer, GNU C Library
On 03/29/2016 07:01 AM, Florian Weimer wrote:
> This is a minor security issue in nss_dns, triggered by a very long name
> passed to getnetbyname.
>
> The defensive copy is not needed because the name may not alias the
> output buffer.
This code dates back to the original 1996 sources that were included in glibc
and I expect that at that point the propagation of `const char *` for name
was not entirely complete and that at some point the internals assumed they
could scribble on name. That is not the case today, we propagate `const char *`
to all callers, and a rather deep review that I just did using ctags shows
that the call tree preserves the constness of name.
This patch looks good to me.
> 2016-03-29 Florian Weimer <fweimer@redhat.com>
>
> [BZ #19879]
> CVE-2016-3075
> * resolv/nss_dns/dns-network.c (_nss_dns_getnetbyname_r): Do not
> copy name.
>
> diff --git a/resolv/nss_dns/dns-network.c b/resolv/nss_dns/dns-network.c
> index 2eb2f67..8f301a7 100644
> --- a/resolv/nss_dns/dns-network.c
> +++ b/resolv/nss_dns/dns-network.c
> @@ -118,17 +118,14 @@ _nss_dns_getnetbyname_r (const char *name, struct netent *result,
> } net_buffer;
> querybuf *orig_net_buffer;
> int anslen;
> - char *qbuf;
> enum nss_status status;
>
> if (__res_maybe_init (&_res, 0) == -1)
> return NSS_STATUS_UNAVAIL;
>
> - qbuf = strdupa (name);
> -
> net_buffer.buf = orig_net_buffer = (querybuf *) alloca (1024);
>
> - anslen = __libc_res_nsearch (&_res, qbuf, C_IN, T_PTR, net_buffer.buf->buf,
> + anslen = __libc_res_nsearch (&_res, name, C_IN, T_PTR, net_buffer.buf->buf,
> 1024, &net_buffer.ptr, NULL, NULL, NULL, NULL);
> if (anslen < 0)
> {
--
Cheers,
Carlos.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] CVE-2016-3075: Stack overflow in _nss_dns_getnetbyname_r [BZ #19879]
2016-03-29 11:01 [PATCH] CVE-2016-3075: Stack overflow in _nss_dns_getnetbyname_r [BZ #19879] Florian Weimer
2016-03-29 13:49 ` Carlos O'Donell
@ 2016-03-29 21:41 ` Joseph Myers
2016-04-29 8:49 ` Florian Weimer
1 sibling, 1 reply; 4+ messages in thread
From: Joseph Myers @ 2016-03-29 21:41 UTC (permalink / raw)
To: Florian Weimer; +Cc: GNU C Library
On Tue, 29 Mar 2016, Florian Weimer wrote:
> This is a minor security issue in nss_dns, triggered by a very long name
> passed to getnetbyname.
As a security issue it should have an entry in the "Security related
changes" section of NEWS for 2.24.
--
Joseph S. Myers
joseph@codesourcery.com
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] CVE-2016-3075: Stack overflow in _nss_dns_getnetbyname_r [BZ #19879]
2016-03-29 21:41 ` Joseph Myers
@ 2016-04-29 8:49 ` Florian Weimer
0 siblings, 0 replies; 4+ messages in thread
From: Florian Weimer @ 2016-04-29 8:49 UTC (permalink / raw)
To: Joseph Myers; +Cc: GNU C Library
On 03/29/2016 11:41 PM, Joseph Myers wrote:
> On Tue, 29 Mar 2016, Florian Weimer wrote:
>
>> This is a minor security issue in nss_dns, triggered by a very long name
>> passed to getnetbyname.
>
> As a security issue it should have an entry in the "Security related
> changes" section of NEWS for 2.24.
I expected to wait with this until closer to the 2.24, but I guess there
is no harm in adding these entries now.
I pushed the following NEWS entry:
* An unnecessary stack copy in _nss_dns_getnetbyname_r was removed. It
could result in a stack overflow when getnetbyname was called with an
overly long name. (CVE-2016-3075)
Thanks,
Florian
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2016-04-29 8:49 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-03-29 11:01 [PATCH] CVE-2016-3075: Stack overflow in _nss_dns_getnetbyname_r [BZ #19879] Florian Weimer
2016-03-29 13:49 ` Carlos O'Donell
2016-03-29 21:41 ` Joseph Myers
2016-04-29 8:49 ` Florian Weimer
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).