Seems to be fine with the patch applied.
Thanks.
On Sun, 2006-03-05 at 19:02 -0500, tony@develop-help.com via RT wrote:
Show quoted text> On Sat, Mar 04, 2006 at 12:31:13PM -0500, Guest via RT wrote:
> > In Bricolage with thumbnails enabled I receive the below apache crash
> > (mod_perl) with Imager 0.47 and later. 0.45 is confirmed to function on
> > these images correctly. Three image examples that work in 0.45 and
> > crash 0.47 and 0.48 are attached.
>
> And so I forgot the patch. Attached this time.
>
> Tony
>
> plain text document attachment (exif_user_comment.diff)
> Index: MANIFEST
> ===================================================================
> --- MANIFEST (revision 933)
> +++ MANIFEST (revision 934)
> @@ -209,6 +209,7 @@
> t/t92samples.t
> t/testtools.pl
> tags.c
> +testimg/209_yonge.jpg Regression test: #17981
> testimg/bad1oflow.bmp 1-bit/pixel, overflow integer on 32-bit machines
> testimg/bad1wid0.bmp 1-bit/pixel, zero width
> testimg/bad24comp.bmp 24-bit/pixel, bad compression
> Index: Changes
> ===================================================================
> --- Changes (revision 933)
> +++ Changes (revision 934)
> @@ -1366,12 +1366,19 @@
> custom META.yml was a waste.
> - bump to 0.47_01
>
> -0.48
> +0.48 Fri 3 Mar 2006
> - removed unused hashinfo() function from Imager.xs
> - added =items for various methods, so Pod::Coverage will pick them up
> (Pod::Coverage tests to be added in 0.49)
> - bump to 0.48
>
> +0.49
> +- handle short EXIF user_comment fields correctly, previously Imager
> + would read (and potentially) write beyond the end of an allocated block,
> + or through a NULL pointer if the EXIF user_comment field was less
> + than 8 bytes long.
> +
https://rt.cpan.org/Ticket/Display.html?id=17981
> +
> =================================================================
>
> For latest versions check the Imager-devel pages:
> Index: imexif.c
> ===================================================================
> --- imexif.c (revision 933)
> +++ imexif.c (revision 934)
> @@ -695,7 +695,7 @@
> memcpy(user_comment, tiff->base + entry->offset, entry->size);
> /* the first 8 bytes indicate the encoding, make them into spaces
> for better presentation */
> - for (i = 0; i < 8; ++i) {
> + for (i = 0; i < entry->size && i < 8; ++i) {
> if (user_comment[i] == '\0')
> user_comment[i] = ' ';
> }
> Index: t/t101jpeg.t
> ===================================================================
> --- t/t101jpeg.t (revision 933)
> +++ t/t101jpeg.t (revision 934)
> @@ -2,7 +2,7 @@
> use strict;
> use lib 't';
> use Imager qw(:all);
> -use Test::More tests => 49;
> +use Test::More tests => 51;
>
> init_log("testout/t101jpeg.log",1);
>
> @@ -30,7 +30,7 @@
> $im = Imager->new(xsize=>2, ysize=>2);
> ok(!$im->write(file=>"testout/nojpeg.jpg"), "should fail to write jpeg");
> cmp_ok($im->errstr, '=~', qr/format not supported/, "check no jpeg message");
> - skip("no jpeg support", 45);
> + skip("no jpeg support", 47);
> }
> } else {
> open(FH,">testout/t101.jpg") || die "cannot open testout/t101.jpg for writing\n";
> @@ -233,5 +233,17 @@
> is_deeply($expect_tags, \%tags, "check tags for $filename");
> }
> }
> +
> + { # Issue # 17981
> + # the test image has a zero-length user_comment field
> + # the code would originally attempt to convert '\0' to ' '
> + # for the first 8 bytes, even if the string was less than
> + # 8 bytes long
> + my $im = Imager->new;
> + ok($im->read(file => 'testimg/209_yonge.jpg', type=>'jpeg'),
> + "test read of image with invalid exif_user_comment");
> + is($im->tags(name=>'exif_user_comment'), '',
> + "check exif_user_comment set correctly");
> + }
> }
>
> Index: testimg/209_yonge.jpg
> ===================================================================
> Cannot display: file marked as a binary type.
> svn:mime-type = application/octet-stream
>
> Property changes on: testimg/209_yonge.jpg
> ___________________________________________________________________
> Name: svn:mime-type
> + application/octet-stream
