All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] complete e390f9a port for v4.9.106
@ 2018-06-09 11:42 Philip Müller
  2018-06-09 14:05 ` Greg Kroah-Hartman
  0 siblings, 1 reply; 4+ messages in thread
From: Philip Müller @ 2018-06-09 11:42 UTC (permalink / raw)
  To: Greg Kroah-Hartman; +Cc: stable, oberon

[-- Attachment #1: Type: text/plain, Size: 1077 bytes --]

objtool ports introduced in v4.9.106 were not totally complete. Therefore
they resulted in issues like:

  module: overflow in relocation type 10 val XXXXXXXXXXX
  ‘usbcore’ likely not compiled with -mcmodel=kernel
  module: overflow in relocation type 10 val XXXXXXXXXXX
  ‘scsi_mod’ likely not compiled with -mcmodel=kernel

Missing part was the complete backport of commit e390f9a.

Original notes by Josh Poimboeuf:

The '__unreachable' and '__func_stack_frame_non_standard' sections are
only used at compile time.  They're discarded for vmlinux but they
should also be discarded for modules.

Since this is a recurring pattern, prefix the section names with
".discard.".  It's a nice convention and vmlinux.lds.h already discards
such sections.

Also remove the 'a' (allocatable) flag from the __unreachable section
since it doesn't make sense for a discarded section.

Signed-off-by: Philip Müller <philm@manjaro.org>
Fixes: d1091c7fa3d5 ("objtool: Improve detection of BUG() and other dead
ends")
Link: https://gitlab.manjaro.org/packages/core/linux49/issues/2

[-- Attachment #2: complete-e390f9a-port.patch --]
[-- Type: text/x-patch, Size: 970 bytes --]

diff -Npur linux-4.9.106.orig/arch/x86/kernel/vmlinux.lds.S linux-4.9.106/arch/x86/kernel/vmlinux.lds.S
--- linux-4.9.106.orig/arch/x86/kernel/vmlinux.lds.S	2018-06-09 12:43:04.238314000 +0200
+++ linux-4.9.106/arch/x86/kernel/vmlinux.lds.S	2018-06-09 12:44:39.034361441 +0200
@@ -352,8 +352,6 @@ SECTIONS
 	DISCARDS
 	/DISCARD/ : {
 		*(.eh_frame)
-		*(__func_stack_frame_non_standard)
-		*(__unreachable)
 	}
 }
 
diff -Npur linux-4.9.106.orig/include/linux/compiler-gcc.h linux-4.9.106/include/linux/compiler-gcc.h
--- linux-4.9.106.orig/include/linux/compiler-gcc.h	2018-06-09 12:43:05.058337000 +0200
+++ linux-4.9.106/include/linux/compiler-gcc.h	2018-06-09 12:45:35.779367605 +0200
@@ -206,7 +206,7 @@
 #ifdef CONFIG_STACK_VALIDATION
 #define annotate_unreachable() ({					\
 	asm("1:\t\n"							\
-	    ".pushsection __unreachable, \"a\"\t\n"			\
+	    ".pushsection .discard.unreachable\t\n"			\
 	    ".long 1b\t\n"						\
 	    ".popsection\t\n");						\
 })

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] complete e390f9a port for v4.9.106
  2018-06-09 11:42 [PATCH] complete e390f9a port for v4.9.106 Philip Müller
@ 2018-06-09 14:05 ` Greg Kroah-Hartman
  2018-06-09 19:35   ` Philip Müller
  0 siblings, 1 reply; 4+ messages in thread
From: Greg Kroah-Hartman @ 2018-06-09 14:05 UTC (permalink / raw)
  To: Philip Müller; +Cc: stable, oberon

On Sat, Jun 09, 2018 at 01:42:05PM +0200, Philip Müller wrote:
> objtool ports introduced in v4.9.106 were not totally complete. Therefore
> they resulted in issues like:
> 
>   module: overflow in relocation type 10 val XXXXXXXXXXX
>   ‘usbcore’ likely not compiled with -mcmodel=kernel
>   module: overflow in relocation type 10 val XXXXXXXXXXX
>   ‘scsi_mod’ likely not compiled with -mcmodel=kernel

What is causing these messages?  What arch and .config?

> Missing part was the complete backport of commit e390f9a.

Ah, right, I missed those other parts of this commit, good catch.

> 
> Original notes by Josh Poimboeuf:
> 
> The '__unreachable' and '__func_stack_frame_non_standard' sections are
> only used at compile time.  They're discarded for vmlinux but they
> should also be discarded for modules.
> 
> Since this is a recurring pattern, prefix the section names with
> ".discard.".  It's a nice convention and vmlinux.lds.h already discards
> such sections.
> 
> Also remove the 'a' (allocatable) flag from the __unreachable section
> since it doesn't make sense for a discarded section.
> 
> Signed-off-by: Philip Müller <philm@manjaro.org>
> Fixes: d1091c7fa3d5 ("objtool: Improve detection of BUG() and other dead
> ends")
> Link: https://gitlab.manjaro.org/packages/core/linux49/issues/2

> diff -Npur linux-4.9.106.orig/arch/x86/kernel/vmlinux.lds.S linux-4.9.106/arch/x86/kernel/vmlinux.lds.S
> --- linux-4.9.106.orig/arch/x86/kernel/vmlinux.lds.S	2018-06-09 12:43:04.238314000 +0200
> +++ linux-4.9.106/arch/x86/kernel/vmlinux.lds.S	2018-06-09 12:44:39.034361441 +0200
> @@ -352,8 +352,6 @@ SECTIONS
>  	DISCARDS
>  	/DISCARD/ : {
>  		*(.eh_frame)
> -		*(__func_stack_frame_non_standard)
> -		*(__unreachable)
>  	}
>  }
>  
> diff -Npur linux-4.9.106.orig/include/linux/compiler-gcc.h linux-4.9.106/include/linux/compiler-gcc.h
> --- linux-4.9.106.orig/include/linux/compiler-gcc.h	2018-06-09 12:43:05.058337000 +0200
> +++ linux-4.9.106/include/linux/compiler-gcc.h	2018-06-09 12:45:35.779367605 +0200
> @@ -206,7 +206,7 @@
>  #ifdef CONFIG_STACK_VALIDATION
>  #define annotate_unreachable() ({					\
>  	asm("1:\t\n"							\
> -	    ".pushsection __unreachable, \"a\"\t\n"			\
> +	    ".pushsection .discard.unreachable\t\n"			\
>  	    ".long 1b\t\n"						\
>  	    ".popsection\t\n");						\
>  })


What about the fix up to scripts/mod/modpost and module-common.lds?  Are
they not needed?

thanks,

greg k-h

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] complete e390f9a port for v4.9.106
  2018-06-09 14:05 ` Greg Kroah-Hartman
@ 2018-06-09 19:35   ` Philip Müller
  2018-06-12 16:16     ` Greg Kroah-Hartman
  0 siblings, 1 reply; 4+ messages in thread
From: Philip Müller @ 2018-06-09 19:35 UTC (permalink / raw)
  To: Greg Kroah-Hartman; +Cc: stable, oberon

On 09.06.2018 16:05, Greg Kroah-Hartman wrote:
> 
> What about the fix up to scripts/mod/modpost and module-common.lds?  Are
> they not needed?
> 
> thanks,
> 
> greg k-h
> 

Hi Greg,

I went thru the whole patch. You already had everything merged, except
those two, I added with my patch.

Best, Philip

---

include/linux/frame.h

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/include/linux/frame.h?h=linux-4.9.y&id=35aee626fa6311c3942fedd50c50f6748922e89f

scripts/mod/modpost.c

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/scripts/mod/modpost.c?h=linux-4.9.y&id=35aee626fa6311c3942fedd50c50f6748922e89f

scripts/module-commons.lds

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/scripts/module-common.lds?h=linux-4.9.y&id=35aee626fa6311c3942fedd50c50f6748922e89f

tools/objtool/builtin-check.c

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/tools/objtool/builtin-check.c?h=linux-4.9.y&id=35aee626fa6311c3942fedd50c50f6748922e89f
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/tools/objtool/builtin-check.c?h=linux-4.9.y&id=24ac7a44f7201fdecb2b4068de1ac808f265ede4

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] complete e390f9a port for v4.9.106
  2018-06-09 19:35   ` Philip Müller
@ 2018-06-12 16:16     ` Greg Kroah-Hartman
  0 siblings, 0 replies; 4+ messages in thread
From: Greg Kroah-Hartman @ 2018-06-12 16:16 UTC (permalink / raw)
  To: Philip Müller; +Cc: stable, oberon

On Sat, Jun 09, 2018 at 09:35:27PM +0200, Philip M�ller wrote:
> On 09.06.2018 16:05, Greg Kroah-Hartman wrote:
> > 
> > What about the fix up to scripts/mod/modpost and module-common.lds?  Are
> > they not needed?
> > 
> > thanks,
> > 
> > greg k-h
> > 
> 
> Hi Greg,
> 
> I went thru the whole patch. You already had everything merged, except
> those two, I added with my patch.

Thanks for the patch, now queued up.

greg k-h

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2018-06-12 16:16 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-06-09 11:42 [PATCH] complete e390f9a port for v4.9.106 Philip Müller
2018-06-09 14:05 ` Greg Kroah-Hartman
2018-06-09 19:35   ` Philip Müller
2018-06-12 16:16     ` Greg Kroah-Hartman

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.