UML fails to build on arch i386

UML fails to build on arch i386

[Ritesh Raj Sarraf]

[-- Attachment #1.1: Type: text/plain, Size: 1648 bytes --]


Hello Richard/UML Team,

In preparing the 4.16 UML build, the code fails to build from source on
i386. It build successfully on arch amd64.

Full build log is available at:
https://buildd.debian.org/status/fetch.php?pkg=user-mode-linux&arch=i38
6&ver=4.16-1um-1&stamp=1528784216&raw=0

  CC      arch/um/kernel/tlb.o
  CC      arch/um/kernel/trap.o
  CC      arch/um/kernel/um_arch.o
  CC      arch/um/kernel/umid.o
  CC      arch/um/kernel/maccess.o
  CC      arch/um/kernel/kmsg_dump.o
  CC      arch/um/kernel/skas/clone.o
In file included from ./arch/x86/um/shared/sysdep/ptrace.h:42:0,
                 from ./arch/um/include/shared/as-layout.h:38,
                 from arch/um/kernel/skas/clone.c:11:
./arch/x86/um/shared/sysdep/ptrace_32.h:16:23: 
                       ^
scripts/Makefile.build:329: recipe for target 'arch/um/kernel/skas/clone.o' failed
make[3]: *** [arch/um/kernel/skas/clone.o] Error 1
scripts/Makefile.build:588: recipe for target 'arch/um/kernel/skas' failed
make[2]: *** [arch/um/kernel/skas] Error 2
Makefile:1048: recipe for target 'arch/um/kernel' failed
make[1]: *** [arch/um/kernel] Error 2
make[1]: Leaving directory '/<<PKGBUILDDIR>>/linux-source-4.16'
debian/rules:74: recipe for target 'build-stamp' failed
make: *** [build-stamp] Error 2
dpkg-buildpackage: error: debian/rules build-arch subprocess returned exit status 2
--------------------------------------------------------------------------------
Build finished at 2018-06-12T06:16:48Z

Finished
--------


-- 
Ritesh Raj Sarraf
RESEARCHUT - http://www.researchut.com
"Necessity is the mother of invention."

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 152 bytes --]

_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

Re: UML fails to build on arch i386

[Ritesh Raj Sarraf]

[-- Attachment #1.1: Type: text/plain, Size: 1686 bytes --]

On Thu, 2018-06-14 at 11:01 +0545, Ritesh Raj Sarraf wrote:
>   CC      arch/um/kernel/skas/clone.o
> In file included from ./arch/x86/um/shared/sysdep/ptrace.h:42:0,
>                  from ./arch/um/include/shared/as-layout.h:38,
>                  from arch/um/kernel/skas/clone.c:11:
> ./arch/x86/um/shared/sysdep/ptrace_32.h:16:23: 
>                        ^

This log missed the main error line:

  CC      arch/um/kernel/maccess.o
  CC      arch/um/kernel/kmsg_dump.o
  CC      arch/um/kernel/skas/clone.o
In file included from ./arch/x86/um/shared/sysdep/ptrace.h:42:0,
                 from ./arch/um/include/shared/as-layout.h:38,
                 from arch/um/kernel/skas/clone.c:11:
./arch/x86/um/shared/sysdep/ptrace_32.h:16:23: error: expected identifier before numeric constant
 #define PTRACE_SYSEMU 31
                       ^
scripts/Makefile.build:329: recipe for target 'arch/um/kernel/skas/clone.o' failed
make[3]: *** [arch/um/kernel/skas/clone.o] Error 1
scripts/Makefile.build:588: recipe for target 'arch/um/kernel/skas' failed
make[2]: *** [arch/um/kernel/skas] Error 2
Makefile:1048: recipe for target 'arch/um/kernel' failed
make[1]: *** [arch/um/kernel] Error 2
make[1]: Leaving directory '/<<PKGBUILDDIR>>/linux-source-4.16'
debian/rules:74: recipe for target 'build-stamp' failed
make: *** [build-stamp] Error 2
dpkg-buildpackage: error: debian/rules build-arch subprocess returned exit status 2
--------------------------------------------------------------------------------
Build finished at 2018-06-12T06:16:48Z


-- 
Ritesh Raj Sarraf
RESEARCHUT - http://www.researchut.com
"Necessity is the mother of invention."

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 152 bytes --]

_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

Re: UML fails to build on arch i386

[Richard Weinberger]

On Thu, Jun 14, 2018 at 10:58 AM, Ritesh Raj Sarraf <rrs@researchut.com> wrote:
> On Thu, 2018-06-14 at 11:01 +0545, Ritesh Raj Sarraf wrote:
>>   CC      arch/um/kernel/skas/clone.o
>> In file included from ./arch/x86/um/shared/sysdep/ptrace.h:42:0,
>>                  from ./arch/um/include/shared/as-layout.h:38,
>>                  from arch/um/kernel/skas/clone.c:11:
>> ./arch/x86/um/shared/sysdep/ptrace_32.h:16:23:
>>                        ^
>
> This log missed the main error line:
>
>   CC      arch/um/kernel/maccess.o
>   CC      arch/um/kernel/kmsg_dump.o
>   CC      arch/um/kernel/skas/clone.o
> In file included from ./arch/x86/um/shared/sysdep/ptrace.h:42:0,
>                  from ./arch/um/include/shared/as-layout.h:38,
>                  from arch/um/kernel/skas/clone.c:11:
> ./arch/x86/um/shared/sysdep/ptrace_32.h:16:23: error: expected identifier before numeric constant
>  #define PTRACE_SYSEMU 31

Hmm, can you please check how PTRACE_SYSEMU is defined in your UAPI?

The code is pretty boring:
#ifndef PTRACE_SYSEMU
#define PTRACE_SYSEMU 31
#endif

My best guess is that PTRACE_SYSEMU is not a define but something else and then
#define fails in an interesting way.

-- 
Thanks,
//richard

_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

Re: UML fails to build on arch i386

[Ritesh Raj Sarraf]

[-- Attachment #1.1: Type: text/plain, Size: 1705 bytes --]

On Thu, 2018-06-14 at 13:15 +0200, Richard Weinberger wrote:
> > This log missed the main error line:
> > 
> >    CC      arch/um/kernel/maccess.o
> >    CC      arch/um/kernel/kmsg_dump.o
> >    CC      arch/um/kernel/skas/clone.o
> > In file included from ./arch/x86/um/shared/sysdep/ptrace.h:42:0,
> >                   from ./arch/um/include/shared/as-layout.h:38,
> >                   from arch/um/kernel/skas/clone.c:11:
> > ./arch/x86/um/shared/sysdep/ptrace_32.h:16:23: error: expected
> > identifier before numeric constant
> >   #define PTRACE_SYSEMU 31
> 
> Hmm, can you please check how PTRACE_SYSEMU is defined in your UAPI?
> 
> The code is pretty boring:
> #ifndef PTRACE_SYSEMU
> #define PTRACE_SYSEMU 31
> #endif
> 
> My best guess is that PTRACE_SYSEMU is not a define but something
> else and then
> #define fails in an interesting way.

There's nothing much in here either.
From file: i386-linux-gnu/sys/ptrace.h
And it is the same code on x86_64 too, where I've been able to
successfully build it.



#ifdef __x86_64__
  /* Access TLS data.  */
  PTRACE_ARCH_PRCTL = 30,
# define PT_ARCH_PRCTL PTRACE_ARCH_PRCTL
#endif

  /* Continue and stop at the next syscall, it will not be executed.  */
  PTRACE_SYSEMU = 31,
#define PT_SYSEMU PTRACE_SYSEMU

  /* Single step the process, the next syscall will not be executed.  */
  PTRACE_SYSEMU_SINGLESTEP = 32,
#define PT_SYSEMU_SINGLESTEP PTRACE_SYSEMU_SINGLESTEP

  /* Execute process until next taken branch.  */
  PTRACE_SINGLEBLOCK = 33,
#define PT_STEPBLOCK PTRACE_SINGLEBLOCK



-- 
Ritesh Raj Sarraf
RESEARCHUT - http://www.researchut.com
"Necessity is the mother of invention."

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 152 bytes --]

_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

Re: UML fails to build on arch i386

[Richard Weinberger]

Am Donnerstag, 14. Juni 2018, 17:44:37 CEST schrieb Ritesh Raj Sarraf:
> On Thu, 2018-06-14 at 13:15 +0200, Richard Weinberger wrote:
> > > This log missed the main error line:
> > > 
> > >    CC      arch/um/kernel/maccess.o
> > >    CC      arch/um/kernel/kmsg_dump.o
> > >    CC      arch/um/kernel/skas/clone.o
> > > In file included from ./arch/x86/um/shared/sysdep/ptrace.h:42:0,
> > >                   from ./arch/um/include/shared/as-layout.h:38,
> > >                   from arch/um/kernel/skas/clone.c:11:
> > > ./arch/x86/um/shared/sysdep/ptrace_32.h:16:23: error: expected
> > > identifier before numeric constant
> > >   #define PTRACE_SYSEMU 31
> > 
> > Hmm, can you please check how PTRACE_SYSEMU is defined in your UAPI?
> > 
> > The code is pretty boring:
> > #ifndef PTRACE_SYSEMU
> > #define PTRACE_SYSEMU 31
> > #endif
> > 
> > My best guess is that PTRACE_SYSEMU is not a define but something
> > else and then
> > #define fails in an interesting way.
> 
> There's nothing much in here either.
> From file: i386-linux-gnu/sys/ptrace.h
> And it is the same code on x86_64 too, where I've been able to
> successfully build it.
> 
> 
> 
> #ifdef __x86_64__
>   /* Access TLS data.  */
>   PTRACE_ARCH_PRCTL = 30,
> # define PT_ARCH_PRCTL PTRACE_ARCH_PRCTL
> #endif
> 
>   /* Continue and stop at the next syscall, it will not be executed.  */
>   PTRACE_SYSEMU = 31,
> #define PT_SYSEMU PTRACE_SYSEMU

Okay, PTRACE_SYSEMU seems to be part of an enum, that's why UML's #ifndef does not
not protect the #define.

What libc is this?

I think we can remove UML's define check and just assume that every system has PTRACE_SYSEMU
properly defined.

Thanks,
//richard

-- 
sigma star gmbh - Eduard-Bodem-Gasse 6 - 6020 Innsbruck - Austria
ATU66964118 - FN 374287y

_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

Re: UML fails to build on arch i386

[Ritesh Raj Sarraf]

[-- Attachment #1.1: Type: text/plain, Size: 1297 bytes --]

Hello Richard,

On Thu, 2018-06-14 at 20:41 +0200, Richard Weinberger wrote:
> > 
> > #ifdef __x86_64__
> >    /* Access TLS data.  */
> >    PTRACE_ARCH_PRCTL = 30,
> > # define PT_ARCH_PRCTL PTRACE_ARCH_PRCTL
> > #endif
> > 
> >    /* Continue and stop at the next syscall, it will not be
> > executed.  */
> >    PTRACE_SYSEMU = 31,
> > #define PT_SYSEMU PTRACE_SYSEMU
> 
> Okay, PTRACE_SYSEMU seems to be part of an enum, that's why UML's
> #ifndef does not
> not protect the #define.
> 
> What libc is this?

This is glibc 2.27 from Debian.

rrs@priyasi:~$ dpkg -S /usr/include/x86_64-linux-gnu/sys/ptrace.h 
libc6-dev:amd64: /usr/include/x86_64-linux-gnu/sys/ptrace.h
12:19 ♒♒♒   ☺ 😄    
rrs@priyasi:~$ apt policy libc6-dev
libc6-dev:
  Installed: 2.27-3
  Candidate: 2.27-3
  Version table:
 *** 2.27-3 900
        900 http://deb.debian.org/debian testing/main amd64 Packages
        500 http://deb.debian.org/debian unstable/main amd64 Packages
        100 /var/lib/dpkg/status
12:19 ♒♒♒   ☺ 😄    



> 
> I think we can remove UML's define check and just assume that every
> system has PTRACE_SYSEMU
> properly defined.
-- 
Ritesh Raj Sarraf
RESEARCHUT - http://www.researchut.com
"Necessity is the mother of invention."

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 152 bytes --]

_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

Re: UML fails to build on arch i386

[Richard Weinberger]

Am Freitag, 15. Juni 2018, 08:35:02 CEST schrieb Ritesh Raj Sarraf:
> Hello Richard,
> 
> On Thu, 2018-06-14 at 20:41 +0200, Richard Weinberger wrote:
> > > 
> > > #ifdef __x86_64__
> > >    /* Access TLS data.  */
> > >    PTRACE_ARCH_PRCTL = 30,
> > > # define PT_ARCH_PRCTL PTRACE_ARCH_PRCTL
> > > #endif
> > > 
> > >    /* Continue and stop at the next syscall, it will not be
> > > executed.  */
> > >    PTRACE_SYSEMU = 31,
> > > #define PT_SYSEMU PTRACE_SYSEMU
> > 
> > Okay, PTRACE_SYSEMU seems to be part of an enum, that's why UML's
> > #ifndef does not
> > not protect the #define.
> > 
> > What libc is this?
> 
> This is glibc 2.27 from Debian.
> 
> rrs@priyasi:~$ dpkg -S /usr/include/x86_64-linux-gnu/sys/ptrace.h 
> libc6-dev:amd64: /usr/include/x86_64-linux-gnu/sys/ptrace.h
> 12:19 ♒♒♒   ☺ 😄    
> rrs@priyasi:~$ apt policy libc6-dev
> libc6-dev:
>   Installed: 2.27-3
>   Candidate: 2.27-3
>   Version table:
>  *** 2.27-3 900
>         900 http://deb.debian.org/debian testing/main amd64 Packages
>         500 http://deb.debian.org/debian unstable/main amd64 Packages
>         100 /var/lib/dpkg/status
> 12:19 ♒♒♒   ☺ 😄    

Okay, as expected, the ptrace SYSEMU request code is now an enum.
Changed by in glibc:
commit c48831d0eebf876d986919baf2d9240a79192837
Author: Dmitry V. Levin <ldv@altlinux.org>
Date:   Wed Dec 27 00:54:38 2017 +0000

    linux/x86: sync sys/ptrace.h with Linux 4.14 [BZ #22433]
    
    Clone generic linux/sys/ptrace.h, add the following x86-specific
    __ptrace_request constants: PTRACE_GET_THREAD_AREA,
    PTRACE_SET_THREAD_AREA, PTRACE_ARCH_PRCTL, PTRACE_SYSEMU,
    PTRACE_SYSEMU_SINGLESTEP, and PTRACE_SINGLEBLOCK.
    
    [BZ #22433]
    * sysdeps/unix/sysv/linux/x86/sys/ptrace.h: New file.

Does this change make UML build on your host?

diff --git a/arch/x86/um/shared/sysdep/ptrace_32.h b/arch/x86/um/shared/sysdep/ptrace_32.h
index b94a108de1dc..ae00d22bce02 100644
--- a/arch/x86/um/shared/sysdep/ptrace_32.h
+++ b/arch/x86/um/shared/sysdep/ptrace_32.h
@@ -10,20 +10,10 @@
 
 static inline void update_debugregs(int seq) {}
 
-/* syscall emulation path in ptrace */
-
-#ifndef PTRACE_SYSEMU
-#define PTRACE_SYSEMU 31
-#endif
-
 void set_using_sysemu(int value);
 int get_using_sysemu(void);
 extern int sysemu_supported;
 
-#ifndef PTRACE_SYSEMU_SINGLESTEP
-#define PTRACE_SYSEMU_SINGLESTEP 32
-#endif
-
 #define UPT_SYSCALL_ARG1(r) UPT_BX(r)
 #define UPT_SYSCALL_ARG2(r) UPT_CX(r)
 #define UPT_SYSCALL_ARG3(r) UPT_DX(r)

Thanks,
//richard

-- 
sigma star gmbh - Eduard-Bodem-Gasse 6 - 6020 Innsbruck - Austria
ATU66964118 - FN 374287y

_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um

Re: UML fails to build on arch i386

[Ritesh Raj Sarraf]

[-- Attachment #1.1: Type: text/plain, Size: 1754 bytes --]

On Fri, 2018-06-15 at 08:43 +0200, Richard Weinberger wrote:
> Okay, as expected, the ptrace SYSEMU request code is now an enum.
> Changed by in glibc:
> commit c48831d0eebf876d986919baf2d9240a79192837
> Author: Dmitry V. Levin <ldv@altlinux.org>
> Date:   Wed Dec 27 00:54:38 2017 +0000
> 
>     linux/x86: sync sys/ptrace.h with Linux 4.14 [BZ #22433]
>     
>     Clone generic linux/sys/ptrace.h, add the following x86-specific
>     __ptrace_request constants: PTRACE_GET_THREAD_AREA,
>     PTRACE_SET_THREAD_AREA, PTRACE_ARCH_PRCTL, PTRACE_SYSEMU,
>     PTRACE_SYSEMU_SINGLESTEP, and PTRACE_SINGLEBLOCK.
>     
>     [BZ #22433]
>     * sysdeps/unix/sysv/linux/x86/sys/ptrace.h: New file.
> 
> Does this change make UML build on your host?
> 
> diff --git a/arch/x86/um/shared/sysdep/ptrace_32.h
> b/arch/x86/um/shared/sysdep/ptrace_32.h
> index b94a108de1dc..ae00d22bce02 100644
> --- a/arch/x86/um/shared/sysdep/ptrace_32.h
> +++ b/arch/x86/um/shared/sysdep/ptrace_32.h
> @@ -10,20 +10,10 @@
>  
>  static inline void update_debugregs(int seq) {}
>  
> -/* syscall emulation path in ptrace */
> -
> -#ifndef PTRACE_SYSEMU
> -#define PTRACE_SYSEMU 31
> -#endif
> -
>  void set_using_sysemu(int value);
>  int get_using_sysemu(void);
>  extern int sysemu_supported;
>  
> -#ifndef PTRACE_SYSEMU_SINGLESTEP
> -#define PTRACE_SYSEMU_SINGLESTEP 32
> -#endif
> -
>  #define UPT_SYSCALL_ARG1(r) UPT_BX(r)
>  #define UPT_SYSCALL_ARG2(r) UPT_CX(r)
>  #define UPT_SYSCALL_ARG3(r) UPT_DX(r)
> 

Thank you Richard.

That does fix the build. I built and verified it on i386 and amd64.

Thanks,
Ritesh

-- 
Ritesh Raj Sarraf
RESEARCHUT - http://www.researchut.com
"Necessity is the mother of invention."

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 152 bytes --]

_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um