From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:60681) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1cieiZ-0003Fh-O4 for qemu-devel@nongnu.org; Tue, 28 Feb 2017 05:09:16 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1cieiW-0007De-Iy for qemu-devel@nongnu.org; Tue, 28 Feb 2017 05:09:15 -0500 Received: from mx1.redhat.com ([209.132.183.28]:36834) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1cieiW-0007D1-Ce for qemu-devel@nongnu.org; Tue, 28 Feb 2017 05:09:12 -0500 Date: Tue, 28 Feb 2017 10:09:07 +0000 From: "Daniel P. Berrange" Message-ID: <20170228100907.GD2762@redhat.com> Reply-To: "Daniel P. Berrange" References: <20170227201456.31814-1-berrange@redhat.com> <739c3a03-fffe-46d0-0ba6-a328a9a18eb2@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <739c3a03-fffe-46d0-0ba6-a328a9a18eb2@redhat.com> Subject: Re: [Qemu-devel] [PATCH] io: ignore case when matching websockets HTTP headers List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Eric Blake Cc: qemu-devel@nongnu.org, "Denis V. Lunev" , Anton Nefedov On Mon, Feb 27, 2017 at 03:20:37PM -0600, Eric Blake wrote: > On 02/27/2017 02:14 PM, Daniel P. Berrange wrote: > > According to RFC7230 Section 3.2, header field name is case-insensitive. > > Convert the header data into all lowercase before doing string matching > > on the headers. > > > > Signed-off-by: Daniel P. Berrange > > --- > > io/channel-websock.c | 14 +++++++++----- > > 1 file changed, 9 insertions(+), 5 deletions(-) > > > > > @@ -223,7 +223,7 @@ static int qio_channel_websock_handshake_process(QIOChannelWebsock *ioc, > > static int qio_channel_websock_handshake_read(QIOChannelWebsock *ioc, > > Error **errp) > > { > > - char *handshake_end; > > + char *handshake_end, *tmp; > > ssize_t ret; > > /* Typical HTTP headers from novnc are 512 bytes, so limiting > > * total header size to 4096 is easily enough. */ > > Drive-by grammar nit: s/easily/easy/ > > > @@ -249,9 +249,13 @@ static int qio_channel_websock_handshake_read(QIOChannelWebsock *ioc, > > } > > } > > > > + for (tmp = (char *)ioc->encinput.buffer; tmp < handshake_end; tmp++) { > > + *tmp = g_ascii_tolower(*tmp); > > + } > > + > > if (qio_channel_websock_handshake_process(ioc, > > (char *)ioc->encinput.buffer, > > - ioc->encinput.offset, > > + handshake_end - (char *)ioc->encinput.buffer, > > I'm not sure why this change is here; nothing else in the patch changed > ioc->encinput.offset. It is a related bug. The ioc->encinput buffer contains upto 4096 bytes, but that data may cover both the header & payload. The handshake_end pointer refers to the end of the header region. So if we just pass encinput.offset, then handshake_process() will be searching part of the payload as well as the header. This might cause it do match the wrong data. I guess I should have made that clear in the commit, or even tput it in a separate commit Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://entangle-photo.org -o- http://search.cpan.org/~danberr/ :|