Skip to contentSkip to conversation

package/ghostscript: fix static build errors

PATCH New Bernd Kuhls 2 participants 4 messages Aug 18, 2019 – Aug 21, 2019

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/2 package/ghostscript: fix static build errors

Patch 0003 fixes a bug which occured with pkgconfig enabled. Patch 0004 fixes the linking error with libidn.

Fixes:
http://autobuild.buildroot.net/results/45a/45a94a494eb2b719ac338d2e734753b11a4a0144/

---
v2: added BUILD_PKGCONFIG=/bin/false (Thomas)
fixed fontconfig/freetype linking error caused by using pkgconfig updated libidn detection (Thomas)

 .../ghostscript/0003-pkgconf-libs-only.patch  | 28 ++++++++++
 .../ghostscript/0004-pkgconfig-libidn.patch   | 53 +++++++++++++++++++
 package/ghostscript/ghostscript.mk            |  3 ++
 3 files changed, 84 insertions(+)
 create mode 100644 package/ghostscript/0003-pkgconf-libs-only.patch
 create mode 100644 package/ghostscript/0004-pkgconfig-libidn.patch
package/ghostscript/0003-pkgconf-libs-only.patch new file +28 -0
new file mode 100644
index 0000000000..8518c3b1f1
@@ -0,0 +1,28 @@
+fix build with pkg-config enabled
+
+Patch sent upstream:
+https://bugs.ghostscript.com/show_bug.cgi?id=701440
+
+Signed-off-by: Bernd Kuhls <[email protected]>
+
+diff -uNr ghostscript-9.27.old/configure.ac ghostscript-9.27/configure.ac
+--- ghostscript-9.27.old/configure.ac	2019-04-04 09:43:14.000000000 +0200
++++ ghostscript-9.27/configure.ac	2019-08-18 11:32:50.163910476 +0200
+@@ -853,7 +853,7 @@
+                 if $PKGCONFIG --exists fontconfig; then
+                         AC_MSG_RESULT(yes)
+                         FONTCONFIG_CFLAGS="$CFLAGS `$PKGCONFIG --cflags fontconfig`"
+-                        FONTCONFIG_LIBS="`$PKGCONFIG --libs fontconfig`"
++                        FONTCONFIG_LIBS="`$PKGCONFIG --libs-only-l fontconfig`"
+                         HAVE_FONTCONFIG=-DHAVE_FONTCONFIG
+                 else
+                         AC_MSG_RESULT(no)
+@@ -996,7 +996,7 @@
+             if $PKGCONFIG --atleast-version=12.0.6 freetype2; then
+                 AC_MSG_RESULT(yes)
+                 FT_CFLAGS="$CFLAGS `$PKGCONFIG --cflags freetype2`"
+-                FT_LIBS="`$PKGCONFIG --libs freetype2`"
++                FT_LIBS="`$PKGCONFIG --libs-only-l freetype2`"
+                 FT_BRIDGE=1
+                 SHARE_FT=1
+             else
package/ghostscript/0004-pkgconfig-libidn.patch new file +53 -0

53 lines not on this page · raw message

package/ghostscript/ghostscript.mk +3 -0
index fefac2aee3..42af344878 100644
@@ -10,6 +10,8 @@ GHOSTSCRIPT_SOURCE = ghostscript-$(GHOSTSCRIPT_VERSION).tar.xz
 GHOSTSCRIPT_LICENSE = AGPL-3.0
 GHOSTSCRIPT_LICENSE_FILES = LICENSE
 # 0001-Fix-cross-compilation-issue.patch
+# 0003-pkgconf-libs-only.patch
+# 0004-pkgconfig-libidn.patch
 GHOSTSCRIPT_AUTORECONF = YES
 GHOSTSCRIPT_DEPENDENCIES = \
 	host-lcms2 \
@@ -33,6 +35,7 @@ endef
 GHOSTSCRIPT_POST_PATCH_HOOKS += GHOSTSCRIPT_REMOVE_LIBS
 
 GHOSTSCRIPT_CONF_ENV = \
+	BUILD_PKGCONFIG=/bin/false \
 	CCAUX="$(HOSTCC)" \
 	CFLAGSAUX="$(HOST_CFLAGS) $(HOST_LDFLAGS)"
 
Signature · 1 line
2.20.1
#1 lore raw

PATCH 2/2 package/ghostscript: add upstream security fix for CVE-2019-10216

---
 package/ghostscript/0005-CVE-2019-10216.patch | 49 +++++++++++++++++++
 1 file changed, 49 insertions(+)
 create mode 100644 package/ghostscript/0005-CVE-2019-10216.patch
package/ghostscript/0005-CVE-2019-10216.patch new file +49 -0
new file mode 100644
index 0000000000..2d624cd17b
@@ -0,0 +1,49 @@
+From: Chris Liddell <[email protected]>
+Date: Fri, 2 Aug 2019 14:18:26 +0000 (+0100)
+Subject: Bug 701394: protect use of .forceput with executeonly
+X-Git-Url: http://git.ghostscript.com/?p=ghostpdl.git;a=commitdiff_plain;h=5b85ddd19a8420a1bd2d5529325be35d78e94234
+
+Bug 701394: protect use of .forceput with executeonly
+
+Fixes CVE-2019-10216
+
+Signed-off-by: Bernd Kuhls <[email protected]>
+---
+
+diff --git a/Resource/Init/gs_type1.ps b/Resource/Init/gs_type1.ps
+index 6c7735b..a039cce 100644
+--- a/Resource/Init/gs_type1.ps
++++ b/Resource/Init/gs_type1.ps
+@@ -118,25 +118,25 @@
+                          ( to be the same as glyph: ) print 1 index //== exec } if
+                    3 index exch 3 index .forceput
+                                                                  % scratch(string) RAGL(dict) AGL(dict) CharStrings(dict) cstring gname
+-                 }
++                 }executeonly
+                  {pop} ifelse
+-               } forall
++               } executeonly forall
+                pop pop
+-             }
++             } executeonly
+              {
+                pop pop pop
+              } ifelse
+-           }
++           } executeonly
+            {
+                                                                % scratch(string) RAGL(dict) AGL(dict) CharStrings(dict) cstring gname
+              pop pop
+            } ifelse
+-         } forall
++         } executeonly forall
+          3 1 roll pop pop
+-     } if
++     } executeonly if
+      pop
+      dup /.AGLprocessed~GS //true .forceput
+-   } if
++   } executeonly if
+ 
+    %% We need to excute the C .buildfont1 in a stopped context so that, if there
+    %% are errors we can put the stack back sanely and exit. Otherwise callers won't
Signature · 1 line
2.20.1
#2 lore raw

PATCH v2 1/2 package/ghostscript: fix static build errors

In reply to Bernd Kuhls

Hello Bernd,

Thanks for coming up with this new version. See for some questions below.

package/ghostscript/0003-pkgconf-libs-only.patch line 1

Quoted 7 lines · On Sun, 18 Aug 2019 12:14:05 +0200 Bernd Kuhls <[email protected]> wrote:
package/ghostscript/0003-pkgconf-libs-only.patch new file +1 -0
new file mode 100644
index 0000000000..8518c3b1f1
@@ -0,0 +1,28 @@
+fix build with pkg-config enabled

This commit description is too short. What is happening exactly ?

Quoted 5 lines
package/ghostscript/0003-pkgconf-libs-only.patch new file +5 -0
+
+Patch sent upstream:
+https://bugs.ghostscript.com/show_bug.cgi?id=701440
+
+Signed-off-by: Bernd Kuhls <[email protected]>

package/ghostscript/0003-pkgconf-libs-only.patch line 16

Quoted 10 lines
package/ghostscript/0003-pkgconf-libs-only.patch new file +10 -0
+
+diff -uNr ghostscript-9.27.old/configure.ac ghostscript-9.27/configure.ac
+--- ghostscript-9.27.old/configure.ac	2019-04-04 09:43:14.000000000 +0200
++++ ghostscript-9.27/configure.ac	2019-08-18 11:32:50.163910476 +0200
+@@ -853,7 +853,7 @@
+                 if $PKGCONFIG --exists fontconfig; then
+                         AC_MSG_RESULT(yes)
+                         FONTCONFIG_CFLAGS="$CFLAGS `$PKGCONFIG --cflags fontconfig`"
+-                        FONTCONFIG_LIBS="`$PKGCONFIG --libs fontconfig`"
++                        FONTCONFIG_LIBS="`$PKGCONFIG --libs-only-l fontconfig`"

There are plenty of other places besides those two fontconfig/freetype cases where PKGCONFIG --libs gets used. Why aren't these other places also a problem ? What is the actual problem ?

Using --libs-only-l is potentially wrong, as it could be missing some important flags needed at link time.

package/ghostscript/0003-pkgconf-libs-only.patch

Quoted 2 lines
package/ghostscript/0003-pkgconf-libs-only.patch new file +1 -0
 GHOSTSCRIPT_CONF_ENV = \
+	BUILD_PKGCONFIG=/bin/false \

A comment here above this would be useful, as it's kind of tricky to understand why we're passing /bin/false here.

Thanks,

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#3 lore raw

PATCH v2 1/2 package/ghostscript: fix static build errors

Am Mon, 19 Aug 2019 14:50:08 +0200 schrieb Thomas Petazzoni:

Quoted 3 lines

There are plenty of other places besides those two fontconfig/freetype cases where PKGCONFIG --libs gets used. Why aren't these other places also a problem ? What is the actual problem ?

Hi,

upstream responded to my patch and will need more time to investigate:
https://bugs.ghostscript.com/show_bug.cgi?id=701439#c2

I think the genconf binary, built for the host, has problems dealing with the output of PKGCONFIG --libs, using --libs-only-l was just a hack to demonstrate that the problem is with genconf and not with configure/make.

For someone like me unfamiliar with the internals of the ghostscript build system I prefer to wait for upstream fixing the bug so marking
http://patchwork.ozlabs.org/patch/1148830/ with "Changes requested" is ok.

But I reopened http://patchwork.ozlabs.org/patch/1148831/ because this patch fixes a CVE unrelated to the build problem.

Regards, Bernd