Re: [PATCH] check-execstack: Permit sysdeps to xfail some libs

[Samuel Thibault <samuel.thibault@ens-lyon.org>]

Hello,

Samuel Thibault, le dim. 24 juin 2018 00:53:19 +0200, a ecrit:
> Samuel Thibault, le mer. 13 juin 2018 22:38:28 +0200, a ecrit:
> > * scripts/check-execstack.awk: Consider `xfail' variable containing a list
> > of libraries whose stack executability is expected.
> > * elf/Makefile ($(objpfx)check-execstack.out): Pass
> > $(check-execstack-xfail) to check-execstack.awk through `xfail' variable.
> > * sysdeps/mach/hurd/i386/Makefile (check-execstack-xfail): Set to ld.so
> > libc.so libpthread.so.
> 
> Ping?

Ping?

This is the last issue of the testsuite when cross-building from Linux,
it'd be really useful to fix it before 2.28 :)

> > ---
> >  elf/Makefile                    |  2 +-
> >  scripts/check-execstack.awk     | 27 ++++++++++++++++++++++-----
> >  sysdeps/mach/hurd/i386/Makefile |  6 ++++++
> >  3 files changed, 29 insertions(+), 6 deletions(-)
> > 
> > diff --git a/elf/Makefile b/elf/Makefile
> > index f221422de3..b55f457d98 100644
> > --- a/elf/Makefile
> > +++ b/elf/Makefile
> > @@ -1094,7 +1094,7 @@ common-generated += $(all-built-dso:$(common-objpfx)%=%.phdr)
> >  $(objpfx)check-execstack.out: $(..)scripts/check-execstack.awk \
> >  			      $(objpfx)execstack-default \
> >  			      $(all-built-dso:=.phdr)
> > -	LC_ALL=C $(AWK) -f $^ > $@; \
> > +	LC_ALL=C $(AWK) -v "xfail=$(check-execstack-xfail)" -f $^ > $@; \
> >  	$(evaluate-test)
> >  generated += check-execstack.out
> >  
> > diff --git a/scripts/check-execstack.awk b/scripts/check-execstack.awk
> > index 21d37e9f47..cd6b30ed3c 100644
> > --- a/scripts/check-execstack.awk
> > +++ b/scripts/check-execstack.awk
> > @@ -6,7 +6,12 @@
> >  # It fails (1) if any did indicate executable stack.
> >  # It fails (2) if the input did not take the expected form.
> >  
> > -BEGIN { result = sanity = 0; default_exec = -1 }
> > +BEGIN {
> > +  result = sanity = 0; default_exec = -1;
> > +  split(xfail, xfails, " ");
> > +  for (x in xfails)
> > +    expected_fails[xfails[x] ".phdr"] = 1;
> > +}
> >  
> >  /^execstack-no$/ { default_exec = 0; next }
> >  /^execstack-yes$/ { default_exec = 1; next }
> > @@ -17,6 +22,10 @@ function check_one(name) {
> >      result = 2;
> >    }
> >  
> > +  n = split(name, parts, "/");
> > +  basename = parts[n];
> > +  expected_fail = basename in expected_fails;
> > +
> >    if (!sanity) {
> >      print name ": *** input did not look like readelf -l output";
> >      result = 2;
> > @@ -24,12 +33,20 @@ function check_one(name) {
> >      if (stack_line ~ /^.*RW .*$/) {
> >        print name ": OK";
> >      } else if (stack_line ~ /^.*E.*$/) {
> > -      print name ": *** executable stack signaled";
> > -      result = result ? result : 1;
> > +      if (expected_fail) {
> > +	print name ": *** executable stack signaled, expected";
> > +      } else {
> > +	print name ": *** executable stack signaled";
> > +	result = result ? result : 1;
> > +      }
> >      }
> >    } else if (default_exec) {
> > -    print name ": *** no PT_GNU_STACK entry";
> > -    result = result ? result : 1;
> > +    if (expected_fail) {
> > +      print name ": *** no PT_GNU_STACK entry, expected";
> > +    } else {
> > +      print name ": *** no PT_GNU_STACK entry";
> > +      result = result ? result : 1;
> > +    }
> >    } else {
> >      print name ": no PT_GNU_STACK but default is OK";
> >    }
> > diff --git a/sysdeps/mach/hurd/i386/Makefile b/sysdeps/mach/hurd/i386/Makefile
> > index 0b509d9499..9a824fadf3 100644
> > --- a/sysdeps/mach/hurd/i386/Makefile
> > +++ b/sysdeps/mach/hurd/i386/Makefile
> > @@ -96,3 +96,9 @@ endif
> >  ifeq ($(subdir),mach)
> >  test-xfail-check-abi-libmachuser = yes
> >  endif
> > +
> > +ifeq ($(subdir),elf)
> > +# We do use nested functions involving creation of trampolines, notably for
> > +# callbacks whose parameters don't permit to get the context parameters.
> > +check-execstack-xfail += ld.so libc.so libpthread.so
> > +endif
> > -- 
> > 2.17.1
> > 
> 
> -- 
> Samuel
> /*
>  * [...] Note that 120 sec is defined in the protocol as the maximum
>  * possible RTT.  I guess we'll have to use something other than TCP
>  * to talk to the University of Mars.
>  * PAWS allows us longer timeouts and large windows, so once implemented
>  * ftp to mars will work nicely.
>  */
> (from /usr/src/linux/net/inet/tcp.c, concerning RTT [retransmission timeout])

-- 
Samuel
<k> faut en profiter, aujourd'hui, les blagues bidon sont à 100 dollars
 -+- #sos-bourse -+-