Patchwork [5/7] psmisc: bump to version 22.16

login
register
mail settings
Submitter Gustavo Zacarias
Date March 29, 2012, 2:25 a.m.
Message ID <1332987903-12378-5-git-send-email-gustavo@zacarias.com.ar>
Download mbox | patch
Permalink /patch/149353/
State Accepted
Headers show

Comments

Gustavo Zacarias - March 29, 2012, 2:25 a.m.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
---
 package/psmisc/psmisc-link-against-libintl.patch |   38 ++++------------------
 package/psmisc/psmisc.mk                         |    6 ++-
 2 files changed, 11 insertions(+), 33 deletions(-)
Peter Korsgaard - March 29, 2012, 10:01 p.m.
>>>>> "Gustavo" == Gustavo Zacarias <gustavo@zacarias.com.ar> writes:

 Gustavo> Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>

 Gustavo> +# Don't force -fstack-protector
 Gustavo> +PSMISC_CONF_OPT = --disable-harden-flags

We have a BR2_TOOLCHAIN_BUILDROOT_USE_SSP, so we shouldn't disable if we
know the toolchain supports it.

Other than that it looks good - Committed, thanks.
Gustavo Zacarias - March 29, 2012, 10:37 p.m.
On 2012-03-29 19:01, Peter Korsgaard wrote:

>>>>>> "Gustavo" == Gustavo Zacarias <gustavo@zacarias.com.ar> writes:
>
>  Gustavo> Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
>
>  Gustavo> +# Don't force -fstack-protector
>  Gustavo> +PSMISC_CONF_OPT = --disable-harden-flags
>
> We have a BR2_TOOLCHAIN_BUILDROOT_USE_SSP, so we shouldn't disable if 
> we
> know the toolchain supports it.
>
> Other than that it looks good - Committed, thanks.

Yes, however the support is incomplete, we aren't copying libssp for 
external toolchains and we're doing --disable-libssp for the internal 
too so it's better to disable it altogether until that's fixed (i recall 
there was a bug or email reporting it).
Regards.
Peter Korsgaard - March 30, 2012, 10:55 a.m.
>>>>> "Gustavo" == Gustavo Zacarias <gustavo@zacarias.com.ar> writes:

Hi,

 Gustavo> +# Don't force -fstack-protector
 Gustavo> +PSMISC_CONF_OPT = --disable-harden-flags
 >> 
 >> We have a BR2_TOOLCHAIN_BUILDROOT_USE_SSP, so we shouldn't disable
 >> if we know the toolchain supports it.
 >> 
 >> Other than that it looks good - Committed, thanks.

 Gustavo> Yes, however the support is incomplete, we aren't copying
 Gustavo> libssp for external toolchains and we're doing
 Gustavo> --disable-libssp for the internal too so it's better to
 Gustavo> disable it altogether until that's fixed (i recall there was a
 Gustavo> bug or email reporting it).

It is true that we don't have SSP support in external toolchains, but
BR2_TOOLCHAIN_BUILDROOT_USE_SSP isn't enabled for those, and for
internal toolchains we do enable the SSP support in uClibc, so those
should afaik be OK.

So I think the commit is OK as is.
Gustavo Zacarias - March 30, 2012, 11:37 a.m.
On 2012-03-30 07:55, Peter Korsgaard wrote:

> It is true that we don't have SSP support in external toolchains, but
> BR2_TOOLCHAIN_BUILDROOT_USE_SSP isn't enabled for those, and for
> internal toolchains we do enable the SSP support in uClibc, so those
> should afaik be OK.
>
> So I think the commit is OK as is.

Shouldn't we push -fstack-protector into CFLAGS in the toolchain infra 
when the option is enabled?
AFAIK if you build stack protector support without using a form of 
-fstack-protector* in CFLAGS then the packages won't use it, right?
And if we leave it to each package to do it then we're just giving a 
false sense of security with that option.
On a side note, avahi is forcing stack protector off too since memorial 
times, maybe that should be revised?
Regards.
Peter Korsgaard - March 30, 2012, 11:52 a.m.
>>>>> "Gustavo" == Gustavo Zacarias <gustavo@zacarias.com.ar> writes:

Hi,

 Gustavo> Shouldn't we push -fstack-protector into CFLAGS in the toolchain infra
 Gustavo> when the option is enabled?
 Gustavo> AFAIK if you build stack protector support without using a form of
 Gustavo> -fstack-protector* in CFLAGS then the packages won't use it, right?
 Gustavo> And if we leave it to each package to do it then we're just giving a
 Gustavo> false sense of security with that option.
 Gustavo> On a side note, avahi is forcing stack protector off too since
 Gustavo> memorial times, maybe that should be revised?

Yes, I also think our SSP support could be improved. Anyone knows /
cares enough of it to do so?

Patch

diff --git a/package/psmisc/psmisc-link-against-libintl.patch b/package/psmisc/psmisc-link-against-libintl.patch
index fca49c6..907e7e7 100644
--- a/package/psmisc/psmisc-link-against-libintl.patch
+++ b/package/psmisc/psmisc-link-against-libintl.patch
@@ -14,44 +14,20 @@  the "autopoint" program, which itself depends on CVS being installed,
 for some strange reasons. That's the reason why we fall back to the
 hacky solution of modifying the Makefile.in file.
 
-Forward-ported Thomas's patch to psmisc 22.14
+Forward-ported Thomas's patch to psmisc 22.16
 from commit f066ed70cd6939838d4057f66798cbc1d972cc73
 
 Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
 
-diff -Nura psmisc-22.14.orig/src/Makefile.in psmisc-22.14/src/Makefile.in
---- psmisc-22.14.orig/src/Makefile.in	2011-10-12 12:12:09.711983726 -0300
-+++ psmisc-22.14/src/Makefile.in	2011-10-12 12:14:01.733350322 -0300
-@@ -70,16 +70,16 @@
- PROGRAMS = $(bin_PROGRAMS)
- am_fuser_OBJECTS = fuser.$(OBJEXT) signals.$(OBJEXT)
- fuser_OBJECTS = $(am_fuser_OBJECTS)
--fuser_LDADD = $(LDADD)
-+fuser_LDADD = $(LDADD) @INTLLIBS@
- am_killall_OBJECTS = killall.$(OBJEXT) signals.$(OBJEXT)
- killall_OBJECTS = $(am_killall_OBJECTS)
+diff -Nura psmisc-22.16.orig/src/Makefile.in psmisc-22.16/src/Makefile.in
+--- psmisc-22.16.orig/src/Makefile.in	2012-03-28 22:27:33.106660358 -0300
++++ psmisc-22.16/src/Makefile.in	2012-03-28 22:28:55.752219349 -0300
+@@ -76,7 +76,7 @@
  killall_DEPENDENCIES =
  am_peekfd_OBJECTS = peekfd.$(OBJEXT)
  peekfd_OBJECTS = $(am_peekfd_OBJECTS)
 -peekfd_LDADD = $(LDADD)
-+peekfd_LDADD = $(LDADD) @INTLLIBS@
++peekfd_LDADD = $(LDADD) @LIBINTL@
  am_prtstat_OBJECTS = prtstat.$(OBJEXT)
  prtstat_OBJECTS = $(am_prtstat_OBJECTS)
--prtstat_LDADD = $(LDADD)
-+prtstat_LDADD = $(LDADD) @INTLLIBS@
- am_pstree_OBJECTS = pstree.$(OBJEXT)
- pstree_OBJECTS = $(am_pstree_OBJECTS)
- pstree_DEPENDENCIES =
-@@ -217,10 +217,10 @@
- 	$(am__append_10)
- fuser_SOURCES = fuser.c comm.h signals.c signals.h i18n.h fuser.h lists.h
- killall_SOURCES = killall.c comm.h signals.c signals.h i18n.h
--killall_LDADD = @SELINUX_LIB@
-+killall_LDADD = @SELINUX_LIB@ @INTLLIBS@
- peekfd_SOURCES = peekfd.c
- pstree_SOURCES = pstree.c comm.h i18n.h
--pstree_LDADD = @TERMCAP_LIB@ @SELINUX_LIB@
-+pstree_LDADD = @TERMCAP_LIB@ @SELINUX_LIB@ @INTLLIBS@
- prtstat_SOURCES = prtstat.c prtstat.h
- BUILT_SOURCES = signames.h
- EXTRA_DIST = signames.c 
+ prtstat_DEPENDENCIES =
diff --git a/package/psmisc/psmisc.mk b/package/psmisc/psmisc.mk
index df116fc..c6d1108 100644
--- a/package/psmisc/psmisc.mk
+++ b/package/psmisc/psmisc.mk
@@ -4,9 +4,11 @@ 
 #
 #############################################################
 
-PSMISC_VERSION = 22.14
-PSMISC_SITE = http://$(BR2_SOURCEFORGE_MIRROR).dl.sourceforge.net/sourceforge/psmisc
+PSMISC_VERSION = 22.16
+PSMISC_SITE = http://$(BR2_SOURCEFORGE_MIRROR).dl.sourceforge.net/project/psmisc/psmisc
 PSMISC_DEPENDENCIES = ncurses $(if $(BR2_NEEDS_GETTEXT_IF_LOCALE),gettext libintl)
+# Don't force -fstack-protector
+PSMISC_CONF_OPT = --disable-harden-flags
 
 # build after busybox, we prefer fat versions while we're at it
 ifeq ($(BR2_PACKAGE_BUSYBOX),y)