Closed Bug 317620 Opened 14 years ago Closed 14 years ago

Land NSS 3.11 on the Mozilla trunk and on 1_8 branch

Categories

(Firefox Build System :: General, defect)

defect
Not set

Tracking

(Not tracked)

RESOLVED FIXED
mozilla1.8.1

People

(Reporter: wtc, Assigned: KaiE)

References

Details

(Keywords: fixed1.8.1)

Attachments

(6 files, 6 obsolete files)

It is time to upgrade the NSS version on the Mozilla
trunk from NSS 3.10.2 to NSS 3.11.

The difficulty of this NSS upgrade lies in the new
freebl shared libraries/DLLs and the associated .chk
files added in NSS 3.11.  We need to search in the
Mozilla source tree for these strings:
"softokn" or "softokn3"
"freebl_hybrid" or "freebl_hybrid_3"
"freebl_pure32" or "freebl_pure32_3"

Then do the following (each quoted string represents
a shared library/DLL and a .chk file with platform-dependent
file name prefix and suffix added)

1. HP-UX 32-bit PA-RISC:
Change "freebl_hybrid_3" to "freebl_32fpu_3".
Change "freebl_pure32_3" to "freebl_32int_3".

2. Solaris 32-bit SPARC:
Change "freebl_hybrid_3" to "freebl_32fpu_3".
Change "freebl_pure32_3" to "freebl_32int_3".
Add "freebl_32int64_3".

3. Solaris 64-bit SPARC:
Add "freebl_64fpu_3" and "freebl_64int_3".

4. All other platforms:
Add "freebl3".
Wan-Teh: Firefox 1.5 will still work with NSS 3.11, correct?
Attached patch Patch v1 (obsolete) — Splinter Review
First version of patch.

Wan-Teh, is it correct to add the new libs to the link command line?

I have not yet cared for the unix platform specific library additions.
Attachment #204183 - Flags: review?(wtchang)
Wolfgang: Firefox 1.5 is using NSS 3.10.2. Right now
we only plan to integrate NSS 3.11 with the Mozilla
trunk, targeting Firefox 2.0.

Since NSS maintains backward compatibility, Firefox
1.5 will still work with NSS 3.11.

Kai: don't add the new libs to the link command line.
The new libs are all dynamically loaded with
PR_LoadLibraryWithFlags calls at run time.
(The "softokn3" library shouldn't be on the
link command line either.  But it's not necessary to
fix that.)
While updating mozilla trunk to nss 3.11 we might want to have the patch from
bug 101249 as well?
Blocks: eccui
When I tried to apply patch v1 to the tip, it did not apply cleanly for security/manager/Makefile.in.  Attached is the revised portion of the second chunk of that patch so it applies cleanly for me.
(In reply to comment #3)
> Right now we only plan to integrate NSS 3.11 with the
> Mozilla trunk, targeting Firefox 2.0.
Firefox 2.0 will be released from the 1.8 branch, Firefox 3.0 from the current trunk, see http://wiki.mozilla.org/Global:1.9_Trunk_1.8_Branch_Plan
You're right Steffen, I think the new plan is to land on MOZILLA_1_8_BRANCH, too.
Summary: Land NSS 3.11 on the Mozilla trunk. → Land NSS 3.11 on the Mozilla trunk and on 1_8 branch
Attachment #204183 - Attachment is obsolete: true
Attachment #204183 - Flags: review?(wtchang)
ccing mento and smfr for pending camino project file changes. please keep us in the loop, thanks!
Attached patch Patch v2 (obsolete) — Splinter Review
Attachment #205191 - Attachment is obsolete: true
Attachment #206406 - Flags: review?(wtchang)
Comment on attachment 206406 [details] [diff] [review]
Patch v2

Kai,

I didn't realize how much work this is until I
reviewed this patch.  Thanks a lot for your help.

Most of this patch is good.  I have some
suggestions.  I and II are especially important.

I. Mac OS X related changes

In three .pbxproj files, you indicated with
question marks that you don't know what the
magic hexadecimal strings for the new
libfreebl3.dylib entries should be.  Please
ask the Mac developers for help.  I did notice
that there is a gap after the libssl3.dylib
entries.  So if I have to guess, I'd use
libssl3.dylib's magic string for libfreebl3.dylib,
and give libssl3.dylib the next magic string.
(By observing the magic strings for the NSS
.dylib's, you should be able to figure out what
the next magic string should be.)

In mozilla/config/prebind-address-table, you added

> 0x0f000000       @executable_path/libldap50.dylib
> 0x10000000       @executable_path/libxpcom_core.dylib
>+0x11000000       @executable_path/libfreebl3.dylib

We shouldn't need to prebind libfreebl3.dylib
because it is always loaded with a PR_LoadLibrary
call.  I believe this is why libnssckbi.dylib, which
is always loaded with PR_LoadLibrary, is not in the prebind-address-table file.  Please ask a Mac developer
to confirm my understanding.  If we need to prebind
libfreebl3.dylib, we should put it right after
libsoftokn3.dylib; we can either assign 0x0b800000 to
libfreebl3.dylib, or we need to adjust the addresses
for all the .dylib's after libfreebl3.dylib.

II. mozilla/directory/c-sdk changes

These changes need to be reviewed and checked in by
the module owners because that directory has restricted
CVS checkin access.  I also found that their
ldapcsdk_50_client_branch seems quite old:
http://lxr.mozilla.org/seamonkey/source/directory/c-sdk/component_versions.mk#27

 27 # NSS - Network Security Services
 28 NSSVERS                 = 3
 29 NSS_RELEASE_TAG         = NSS_3_3_2_RTM
 30 #NSS_DYNAMIC_SOFTOKN    = 1

Note that NSS_DYNAMIC_SOFTOKN is not set.  This implies
that the parts of the mozilla/directory/c-sdk makefiles
you modified in this patch cannot possibly be used by the
Mozilla clients (because they aren't doing anything with
libsoftokn3.so, which is in the current NSS), and you
should be able to build Mozilla clients without any
changes to mozilla/directory/c-sdk.  Please let me know
if you can build Mozilla clients without changing
mozilla/directory/c-sdk.

It seems that mozilla/directory/c-sdk is designed
so that one can build it with various versions of
NSS.  So, if we need to modify mozilla/directory/c-sdk,
we should ask the module owners if they want our
patch to preserve that.

III. Other changes

Your change to mozilla/nsprpub/pkg/linux/sun-nspr.spec
should be removed from this patch.  Please send an email
to Christophe Ravel at Sun about this minor problem.

In mozilla/security/manager/Makefile.in:

The variable name HAVE_FREEBL_MODULES_3264 looks strange,
but it's just a cosmetic issue.

>+ifeq ($(OS_ARCH), HP-UX)
>+  HAVE_FREEBL_MODULES_32 = 1
>+else
>+  ifneq ($(OS_ARCH), SunOS)
>+    HAVE_FREEBL_MODULES = 1
>+  else
>+    ifeq ($(OS_TEST),i86pc)
>+      HAVE_FREEBL_MODULES = 1
>+    else
>+      ifeq ($(USE_64), 1)
>+        HAVE_FREEBL_MODULES_64 = 1
>+      else
>+        HAVE_FREEBL_MODULES_32 = 1
>+        HAVE_FREEBL_MODULES_3264 = 1
>+      endif
>+    endif
>+  endif
> endif

Rewrite the above as follows (fix the indentation
as you see fit):

# Default
HAVE_FREEBL_MODULES = 1

# 32-bit HP-UX PA-RISC
ifeq ($(OS_ARCH), HP-UX)
ifneq ($(OS_TEST), ia64)
ifndef HAVE_64BIT_OS
HAVE_FREEBL_MODULES =
HAVE_FREEBL_MODULES_32 = 1
endif
endif
endif

# SunOS SPARC
ifeq ($(OS_ARCH), SunOS)
ifneq ($(OS_TEST), i86pc)
ifdef HAVE_64BIT_OS
HAVE_FREEBL_MODULES =
HAVE_FREEBL_MODULES_64 = 1
else
HAVE_FREEBL_MODULES =
HAVE_FREEBL_MODULES_32 = 1
HAVE_FREEBL_MODULES_3264 = 1
endif
endif
endif

Although this format is longer, it is easier to add
a new platform to this format.  Also note that I made
the HP-UX case more specific (32-bit PA-RISC) and I
use HAVE_64BIT_OS instead of USE_64.

>+ifdef HAVE_FREEBL_MODULES
>+FREEBL_LIB = $(DLL_PREFIX)freebl3$(DLL_SUFFIX)
>+FREEBL_CHK = $(DLL_PREFIX)freebl3.chk
> endif

For consistency, please rename FREEBL_LIB as
FREEBL_MODULE.

In mozilla/xpinstall/packager/packages-static-unix:

> ...
> bin/libsoftokn3.chk
> bin/libsoftokn3.so
>+bin/libfreebl3.chk
>+bin/libfreebl3.so
> bin/libssl3.so
> bin/chrome/pipnss.jar
> bin/chrome/pippki.jar
> ; These are required for solaris 32-bit builds
>-bin/libfreebl_pure32_3.chk
>-bin/libfreebl_pure32_3.so
>-bin/libfreebl_hybrid_3.chk
>-bin/libfreebl_hybrid_3.so
>+bin/libfreebl_32fpu_3.chk
>+bin/libfreebl_32fpu_3.so
> ...

Please remove the comment and list all the NSS files
consecutively.

In mozilla/xpinstall/packager/packages-unix:

> ...
> bin/libssl3.so
> bin/chrome/pipnss.jar
> bin/chrome/pippki.jar
>-; These are required for solaris 32-bit builds
>-bin/libfreebl_pure32_3.chk
>-bin/libfreebl_pure32_3.so
>-bin/libfreebl_hybrid_3.chk
>-bin/libfreebl_hybrid_3.so
>+bin/libfreebl3.chk
>+bin/libfreebl3.so
>+bin/libfreebl_32fpu_3.chk
>+bin/libfreebl_32fpu_3.so
> ...

Please list all the NSS files consecutively.

In mozilla/xulrunner/installer/mac/Makefile.in:

> SOFTOKN       = $(BINARY_DIR)/$(DLL_PREFIX)softokn3$(DLL_SUFFIX)
>-FREEBL_HYBRID = $(BINARY_DIR)/$(DLL_PREFIX)freebl_hybrid_3$(DLL_SUFFIX)
>-FREEBL_PURE   = $(BINARY_DIR)/$(DLL_PREFIX)freebl_pure32_3$(DLL_SUFFIX)
>+FREEBL        = $(BINARY_DIR)/$(DLL_PREFIX)freebl3$(DLL_SUFFIX)
>+FREEBL_32FPU  = $(BINARY_DIR)/$(DLL_PREFIX)freebl_32fpu_3$(DLL_SUFFIX)
>+FREEBL_32INT  = $(BINARY_DIR)/$(DLL_PREFIX)freebl_32int_3$(DLL_SUFFIX)

Remove the FREEBL_32FPU and FREEBL_32INT definitions.
They are not used on the Mac, so we can use their default
definitions in $(topsrcdir)/toolkit/mozapps/installer/package.mk,
which this makefile includes.
Attachment #206406 - Flags: review?(wtchang) → review-
(In reply to comment #10)
> (From update of attachment 206406 [details] [diff] [review] [edit])
> Kai,
> I. Mac OS X related changes
> 
> In three .pbxproj files, you indicated with
> question marks that you don't know what the
> magic hexadecimal strings for the new
> libfreebl3.dylib entries should be.  Please
> ask the Mac developers for help.  I did notice
> that there is a gap after the libssl3.dylib
> entries.  So if I have to guess, I'd use
> libssl3.dylib's magic string for libfreebl3.dylib,
> and give libssl3.dylib the next magic string.

Don't do this by hand. You'll have to change the projects in Xcode to get this right. Hop onto #camino for help.
Mike, Simon:

What is the purpose of filess
  mozilla/camino/Camino.xcode/project.pbxproj
  mozilla/embedding/browser/powerplant/PPEmbed.pbproj/project.pbxproj
?


A) for linking the binary? 
or
B) listing the files that should be packaged and shipped?
they are the XCode IDE project files, and serve both purposes (linking and packaging)
> In three .pbxproj files, you indicated with
> question marks that you don't know what the
> magic hexadecimal strings for the new
> libfreebl3.dylib entries should be.  Please
> ask the Mac developers for help.  

I have removed these for now.
Will ask Mac developers for help.


> We shouldn't need to prebind libfreebl3.dylib
> because it is always loaded with a PR_LoadLibrary
> call.  

It sounds reasonable that we do not need to prebind
the freebl library. I have removed the change for now.
Will ask Mac developers for help.


> Your change to mozilla/nsprpub/pkg/linux/sun-nspr.spec
> should be removed from this patch.  Please send an email
> to Christophe Ravel at Sun about this minor problem.

It was an accident that I had this change in my tree.
Removed this change.


> Rewrite the above as follows (fix the indentation
> as you see fit):

done


> For consistency, please rename FREEBL_LIB as
> FREEBL_MODULE.

done


> In mozilla/xpinstall/packager/packages-static-unix:
> Please remove the comment and list all the NSS files
> consecutively.

done


> In mozilla/xpinstall/packager/packages-unix:
> Please list all the NSS files consecutively.

done


> In mozilla/xulrunner/installer/mac/Makefile.in:
> Remove the FREEBL_32FPU and FREEBL_32INT definitions.

done


Now looking into directory/c-sdk
Mike, Simon:

When landing this patch, two new files will be produced (library libfreebl3.dylib + special file libfreebl3.chk).

With NSS 3.11, if these files are not present in the application directory, it will not start up, but will fail with an error message "can not initialize security component".

I think that means the tests on Tinderbox will no longer run successfully.

Therfore we should land the Mac project file changes at the same time.

Simon, do you have a working environment with XCode?

Could you please help us to produce a patch that adds these files to 
  mozilla/camino/Camino.xcode/project.pbxproj
  mozilla/embedding/browser/powerplant/PPEmbed.pbproj/project.pbxproj
  mozilla/embedding/tests/cocoaEmbed/CocoaEmbed.pbproj/project.pbxproj

? Thanks.
> Note that NSS_DYNAMIC_SOFTOKN is not set.  This implies
> that the parts of the mozilla/directory/c-sdk makefiles
> you modified in this patch cannot possibly be used by the
> Mozilla clients (because they aren't doing anything with
> libsoftokn3.so, which is in the current NSS), and you
> should be able to build Mozilla clients without any
> changes to mozilla/directory/c-sdk.  Please let me know
> if you can build Mozilla clients without changing
> mozilla/directory/c-sdk.

Wan-Teh, Mozilla builds fine without the changes to directory/c-sdk. Therefore I have removed those changes from the patch.

(I will send the c-sdk patch portion to the owners of that directory, so they can decide whether they want the patch)
Attached patch Patch v3 (obsolete) — Splinter Review
- addressed review comments as explained above
- removed c-sdk changes, they are not necessary
- integrated patch from bug 288647
Attachment #206406 - Attachment is obsolete: true
Attachment #206920 - Flags: review?(wtchang)
(In reply to comment #15)
> Mike, Simon:
> 
> When landing this patch, two new files will be produced (library
> libfreebl3.dylib + special file libfreebl3.chk).
> 
> With NSS 3.11, if these files are not present in the application directory, it
> will not start up, but will fail with an error message "can not initialize
> security component".

Do we really need the .chk file? We don't package libsoftokn3.chk now (maybe that's a bug).
> Do we really need the .chk file? We don't package libsoftokn3.chk now (maybe
> that's a bug).

Yes we do, and yes that's a bug.

Mozilla's internal crypto component can operate in two different modes.

The default, standard mode, which does not require the .chk file.

A special FIPS mode, that can be enabled using preferences or by using edit/prefs/security devices/enable FIPS.

To my knowledge, all members of US government agencies are required to use FIPS. So if you want to allow them to use your product, you should ship the .chk files.
The CocoaEmbed and PowerPlant projects have bitrotted already (and are not part of the build) so let's leave them for now.
Comment on attachment 206920 [details] [diff] [review]
Patch v3

Kai, this patch is good.  There is only one real
problem.  But I will also suggest a few style changes.

1. mozilla/security/manager/Makefile.in AND
   mozilla/security/manager/psm.pkg

Optional: change "3264" to "32INT64".

Optional: change "MODULE" to "LIB".  For example,
  HAVE_FREEBL_MODULES  ->  HAVE_FREEBL_LIBS
  FREEBL_MODULE        ->  FREEBL_LIB

2. mozilla/xpinstall/packager/packages-static-unix AND
   mozilla/xpinstall/packager/packages-unix

Please list the bin/libfreebl* files below bin/libsoftokn3.so
because they are closely related.

3. mozilla/xulrunner/installer/mac/Makefile.in

This is the only real problem in this patch.  Please add
the FREEBL definition (but not the FREEBL_32FPU and
FREEBL_32INT definitions) in patch v2 (attachment 206406 [details] [diff] [review])
back.
Attachment #206920 - Flags: review?(wtchang) → review-
Attached patch Patch v4 (obsolete) — Splinter Review
> Optional: change "3264" to "32INT64".

done

> Optional: change "MODULE" to "LIB".  For example,
>   HAVE_FREEBL_MODULES  ->  HAVE_FREEBL_LIBS
>   FREEBL_MODULE        ->  FREEBL_LIB

done

> Please list the bin/libfreebl* files below bin/libsoftokn3.so
> because they are closely related.

done

> This is the only real problem in this patch.  Please add
> the FREEBL definition (but not the FREEBL_32FPU and
> FREEBL_32INT definitions) in patch v2 (attachment 206406 [details] [diff] [review] [edit])
> back.

done:

 SOFTOKN       = $(BINARY_DIR)/$(DLL_PREFIX)softokn3$(DLL_SUFFIX)
-FREEBL_HYBRID = $(BINARY_DIR)/$(DLL_PREFIX)freebl_hybrid_3$(DLL_SUFFIX)
-FREEBL_PURE   = $(BINARY_DIR)/$(DLL_PREFIX)freebl_pure32_3$(DLL_SUFFIX)
+FREEBL        = $(BINARY_DIR)/$(DLL_PREFIX)freebl3$(DLL_SUFFIX)
Assignee: wtchang → kengert
Attachment #206920 - Attachment is obsolete: true
Status: NEW → ASSIGNED
Attachment #207990 - Flags: review?(wtchang)
Attached patch Patch v5Splinter Review
This new patch revision contains the change requested by Wan-Teh in bug 288647 comment 48.
Attachment #207990 - Attachment is obsolete: true
Attachment #207994 - Flags: review?(wtchang)
Attachment #207990 - Flags: review?(wtchang)
Comment on attachment 207994 [details] [diff] [review]
Patch v5

r=wtc.  Benjamin, could you please review this patch?
The following is a summary of the changes in this patch.

Most of this patch deals with the new files in NSS 3.11.
- On most platforms, there are two new files:
  libfreebl3.so/libfreebl3.dylib/freebl3.dll and
  libfreebl3.chk/freebl3.chk.
- On Solaris SPARC and 32-bit HP-UX PA-RISC, there are
  more new files.  They all have "freebl3" in their
  file names.
- The *freebl3*.chk files are generated from the corresponding
  *freebl3*.so/.dylib/.dll files by running the "shlibsign"
  tool.

In mozilla/security/manager/Makefile.in, this patch also
contains the fix for bug 288647 to support building NSS
with system NSPR.  Look for the changes related to
ABS_DIST, NSPR_INCLUDE_DIR, and NSPR_LIB_DIR.
Attachment #207994 - Flags: superreview?(benjamin)
Attachment #207994 - Flags: review?(wtchang)
Attachment #207994 - Flags: review+
Blocks: 316714
Depends on: 323079
Attachment #207994 - Flags: superreview?(benjamin) → superreview+
The CVS checkin messages for "Patch v5" (attachment 207994 [details] [diff] [review])
and "Patch for Camino project on trunk" (attachment 206945 [details] [diff] [review])
on the Mozilla trunk (Gecko 1.9 alpha).
Attachment #207994 - Flags: approval1.8.1?
Attachment #206946 - Flags: approval1.8.1?
Since these patches have landed I've not been able to complete a debug build on Windows using MSYS. I've posted the log at:
http://gavinsharp.com/tmp/log.xgt

The fatal error is:
LINK : fatal error LNK1104: cannot open file '\moz\mozilla\ff\dist\lib\secutil.lib'
make[7]: *** [C:/moz-debug/mozilla/ff/nss/freebl/WIN95_SINGLE_SHLIB/freebl3.dll] Error 80
Attachment #208393 - Attachment description: Checked in "Patch v5" and "Patch for Camino project on trunk" on the Mozilla trunk → CVS checkin versions of "Patch v5" and "Patch for Camino project on trunk" on the Mozilla trunk
> cannot open file '\moz\mozilla\ff\dist\lib\secutil.lib'

I don't have a windows system, but at least on Linux I still have libsecutil.a in dist/lib
(In reply to comment #28)
> I don't have a windows system, but at least on Linux I still have libsecutil.a
> in dist/lib

I do too, strangely enough. It doesn't appear to be a permissions issue either. Maybe it is an MSYS issue?
I just learned what MSYS is, I wasn't aware before, it seems to be a secondary way to build on Windows.

Wan-Teh, FYI: http://developer.mozilla.org/en/docs/Building_on_Windows_with_MSYS
MSYS build problems are not considered blockers, it is a tier-3 build configuration.
Not an MSYS problem, apparently. Bug 323645 filed.
*** Bug 323645 has been marked as a duplicate of this bug. ***
The win32 build problem is a combination of factors: using msys, with a non-standard mount point and MSVC.  See bug 323645 for more info.  This patch should fix the problem.
Attachment #208691 - Flags: review?(wtchang)
Comment on attachment 208691 [details] [diff] [review]
Set ABS_DIST to win32 path for msys

r=wtc.  

We should add the -a option to cygpath to more
closely match the configure.in code that sets
MOZ_BUILD_ROOT. 

configure.in also has special code for OS/2 VACPP that
uses "pwd -D" to define MOZ_BUILD_ROOT.  Should we
have the equivalent code here?  I know we only support
OS/2 GCC builds now.
Attachment #208691 - Flags: review?(wtchang) → review+
My trunk Seamonkey build attempt on FreeBSD5.4 ends like this the past
couple of days (I did not try during the period 13-15 Jan):

cd mangle; /usr/local/bin/gmake -j1 libs
FreeBSD5.4_OPT.OBJ/shlibsign -v -i
/scratch/glennrp/Mozilla/Mozilla-1.9a1-060117/mozilla/dist/lib/libsoftokn3.so
Bus error (core dumped)
Comment on attachment 208691 [details] [diff] [review]
Set ABS_DIST to win32 path for msys

I checked in this patch by cls on the Mozilla trunk
to fix the MSYS build failure.

Checking in Makefile.in;
/cvsroot/mozilla/security/manager/Makefile.in,v  <--  Makefile.in
new revision: 1.62; previous revision: 1.61
done
I opened a new bug, #323977, about the trouble on FreeBSD (comment #36)
Attachment #208989 - Flags: review?(wtchang) → review+
Chris, what problem are you trying to avoid when not using the .chk files in the cross compile (is it that they don't exist?).

The .chk files are required on all platforms if FIPS is ever turned on. They are platform independent and can be generated by in either the native or the cross environment (that is any platform can generate valid .chk files for any other platform as long is it has access to the shared libraries the .chk files validate).

bob
It's simply that they don't exist.  My understanding is that in order to build the .chk files, we would have to build complete host versions of DBM, NSPR & NSS to build a host version of shlibgen and we're not setup to do this.
Chris,

Your understanding is correct - local host versions of NSPR/NSS are required in order to make chk files for the target platform. The chk files are a required part of NSS, so this needs to be added for the cross-compile build to be considered complete.
A better (but harder) solution is to ship the shlibsign
tool and use it to generate the .chk files on the target
platform during installation.

Many NSS-based applications "post-process" the NSS libraries
(for example, to strip the libraries of debug information or
to rebase the libraries for faster application startup), so
they already need to run shlibsign (during the build, on the
host platform) to regenerate the .chk files.
Comment on attachment 208989 [details] [diff] [review]
Do not attempt to install .chk files when cross-compiling

I was about to write a patch just like this.  We need this for Mac cross build/universal binary support (bug 322578).
The only problem I see with bundling shlibsign is that most of the post-processing occurs during the product's build & packaging cycle, not their install cycle.  For embedders like minimo, the additional step to sign the libraries during the install phase could be restrictive.  I checked in the patch to hold us over until we can figure this out.

Checking in security/manager/Makefile.in;
/cvsroot/mozilla/security/manager/Makefile.in,v  <--  Makefile.in
new revision: 1.63; previous revision: 1.62
done
Comment on attachment 207994 [details] [diff] [review]
Patch v5

I have no clue who should be making this decision, let me ask drivers@mo.
Attachment #207994 - Flags: approval1.8.1? → branch-1.8.1?
Attachment #206946 - Flags: approval1.8.1? → branch-1.8.1?
Depends on: 325408
Attachment #206946 - Attachment is obsolete: true
Attachment #206946 - Flags: branch-1.8.1?
Note that this patch does not contain any change to the
Camino project, as Mark said in bug 325408 comment 5.
The patch for MOZILLA_1_8_BRANCH was checked in this
afternoon.
Status: ASSIGNED → RESOLVED
Closed: 14 years ago
Keywords: fixed1.8.1
Resolution: --- → FIXED
Target Milestone: --- → mozilla1.8.1
Attachment #207994 - Flags: branch-1.8.1?
Attachment #210653 - Flags: branch-1.8.1+
Depends on: 328522
Product: Core → Firefox Build System
You need to log in before you can comment on or make changes to this bug.