[PATCH 1/22] UML - Include the linker script rather than symlink it

[PATCH 1/22] UML - Include the linker script rather than symlink it

[Jeff Dike]

>From Al Viro:

	Make vmlinux.lds.S include appopriate script instead of playing
games with symlinks.

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Jeff Dike <jdike@addtoit.com>

Index: linux-2.6.11-mm/arch/um/Makefile
===================================================================
--- linux-2.6.11-mm.orig/arch/um/Makefile	2005-04-30 12:57:43.000000000 -0400
+++ linux-2.6.11-mm/arch/um/Makefile	2005-04-30 12:58:23.000000000 -0400
@@ -96,15 +96,11 @@
 
 $(shell cd $(ARCH_DIR) && ln -sf Kconfig_$(SUBARCH) Kconfig_arch)
 
-prepare: $(ARCH_SYMLINKS) $(SYS_HEADERS) $(GEN_HEADERS) \
-	$(ARCH_DIR)/kernel/vmlinux.lds.S
+prepare: $(ARCH_SYMLINKS) $(SYS_HEADERS) $(GEN_HEADERS)
 
 LINK-$(CONFIG_LD_SCRIPT_STATIC) += -static
 LINK-$(CONFIG_LD_SCRIPT_DYN) += -Wl,-rpath,/lib
 
-LD_SCRIPT-$(CONFIG_LD_SCRIPT_STATIC) := uml.lds.S
-LD_SCRIPT-$(CONFIG_LD_SCRIPT_DYN) := dyn.lds.S
-
 CPP_MODE-$(CONFIG_MODE_TT) := -DMODE_TT
 CONFIG_KERNEL_STACK_ORDER ?= 2
 STACK_SIZE := $(shell echo $$[ 4096 * (1 << $(CONFIG_KERNEL_STACK_ORDER)) ] )
@@ -145,15 +141,6 @@
 	@find . \( -name '*.bb' -o -name '*.bbg' -o -name '*.da' \
 		-o -name '*.gcov' \) -type f -print | xargs rm -f
 
-#We need to re-preprocess this when the symlink dest changes.
-#So we touch it when needed.
-$(ARCH_DIR)/kernel/vmlinux.lds.S: FORCE
-	$(Q)if [ "$(shell readlink $@)" != "$(LD_SCRIPT-y)" ]; then \
-		echo '  SYMLINK $@'; \
-		ln -sf $(LD_SCRIPT-y) $@; \
-		touch $@; \
-	fi;
-
 $(SYMLINK_HEADERS):
 	@echo '  SYMLINK $@'
 	$(Q)cd $(TOPDIR)/$(dir $@) ; \
Index: linux-2.6.11-mm/arch/um/kernel/Makefile
===================================================================
--- linux-2.6.11-mm.orig/arch/um/kernel/Makefile	2005-04-30 12:57:43.000000000 -0400
+++ linux-2.6.11-mm/arch/um/kernel/Makefile	2005-04-30 12:59:28.000000000 -0400
@@ -4,7 +4,6 @@
 #
 
 extra-y := vmlinux.lds
-clean-files := vmlinux.lds.S
 
 obj-y = checksum.o config.o exec_kern.o exitcode.o \
 	helper.o init_task.o irq.o irq_user.o ksyms.o main.o mem.o mem_user.o \
Index: linux-2.6.11-mm/arch/um/kernel/vmlinux.lds.S
===================================================================
--- linux-2.6.11-mm.orig/arch/um/kernel/vmlinux.lds.S	2005-04-30 12:56:25.000000000 -0400
+++ linux-2.6.11-mm/arch/um/kernel/vmlinux.lds.S	2005-04-30 12:58:23.000000000 -0400
@@ -1,113 +1,6 @@
-#include <asm-generic/vmlinux.lds.h>
-
-OUTPUT_FORMAT(ELF_FORMAT)
-OUTPUT_ARCH(ELF_ARCH)
-ENTRY(_start)
-jiffies = jiffies_64;
-
-SECTIONS
-{
-  /*This must contain the right address - not quite the default ELF one.*/
-  PROVIDE (__executable_start = START);
-  . = START + SIZEOF_HEADERS;
-
-  /* Used in arch/um/kernel/mem.c. Any memory between START and __binary_start
-   * is remapped.*/
-  __binary_start = .;
-#ifdef MODE_TT
-  .thread_private : {
-    __start_thread_private = .;
-    errno = .;
-    . += 4;
-    arch/um/kernel/tt/unmap_fin.o (.data)
-    __end_thread_private = .;
-  }
-  . = ALIGN(4096);
-  .remap : { arch/um/kernel/tt/unmap_fin.o (.text) }
-
-  /* We want it only if we are in MODE_TT. In both cases, however, when MODE_TT
-   * is off the resulting binary segfaults.*/
-
-  . = ALIGN(4096);		/* Init code and data */
+#include <linux/config.h>
+#ifdef CONFIG_LD_SCRIPT_STATIC
+#include "uml.lds.S"
+#else
+#include "dyn.lds.S"
 #endif
-
-  _stext = .;
-  __init_begin = .;
-  .init.text : {
-	_sinittext = .;
-	*(.init.text)
-	_einittext = .;
-  }
-  . = ALIGN(4096);
-
-  .text      :
-  {
-    *(.text)
-    SCHED_TEXT
-    LOCK_TEXT
-    *(.fixup)
-    /* .gnu.warning sections are handled specially by elf32.em.  */
-    *(.gnu.warning)
-    *(.gnu.linkonce.t*)
-
-    . = ALIGN(4096);
-    __syscall_stub_start = .;
-    *(.__syscall_stub*)
-    __syscall_stub_end = .;
-    . = ALIGN(4096);
-  }
-
-  #include "asm/common.lds.S"
-
-  init.data : { *(init.data) }
-  .data    :
-  {
-    . = ALIGN(KERNEL_STACK_SIZE);		/* init_task */
-    *(.data.init_task)
-    *(.data)
-    *(.gnu.linkonce.d*)
-    CONSTRUCTORS
-  }
-  .data1   : { *(.data1) }
-  .ctors         :
-  {
-    *(.ctors)
-  }
-  .dtors         :
-  {
-    *(.dtors)
-  }
-
-  .got           : { *(.got.plt) *(.got) }
-  .dynamic       : { *(.dynamic) }
-  /* We want the small data sections together, so single-instruction offsets
-     can access them all, and initialized data all before uninitialized, so
-     we can shorten the on-disk segment size.  */
-  .sdata     : { *(.sdata) }
-  _edata  =  .;
-  PROVIDE (edata = .);
-  . = ALIGN(0x1000);
-  .sbss      :
-  {
-   __bss_start = .;
-   PROVIDE(_bss_start = .);
-   *(.sbss)
-   *(.scommon)
-  }
-  .bss       :
-  {
-   *(.dynbss)
-   *(.bss)
-   *(COMMON)
-  }
-  _end = . ;
-  PROVIDE (end = .);
-  /* Stabs debugging sections.  */
-  .stab 0 : { *(.stab) }
-  .stabstr 0 : { *(.stabstr) }
-  .stab.excl 0 : { *(.stab.excl) }
-  .stab.exclstr 0 : { *(.stab.exclstr) }
-  .stab.index 0 : { *(.stab.index) }
-  .stab.indexstr 0 : { *(.stab.indexstr) }
-  .comment 0 : { *(.comment) }
-}

Re: [PATCH 1/22] UML - Include the linker script rather than symlink it

[Andrew Morton]

Jeff Dike <jdike@addtoit.com> wrote:
>
> >From Al Viro:
> 
> 	Make vmlinux.lds.S include appopriate script instead of playing
> games with symlinks.
> 
> Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
> Signed-off-by: Jeff Dike <jdike@addtoit.com>

Confused.

> Index: linux-2.6.11-mm/arch/um/kernel/vmlinux.lds.S
> ===================================================================
> --- linux-2.6.11-mm.orig/arch/um/kernel/vmlinux.lds.S	2005-04-30 12:56:25.000000000 -0400
> +++ linux-2.6.11-mm/arch/um/kernel/vmlinux.lds.S	2005-04-30 12:58:23.000000000 -0400
> @@ -1,113 +1,6 @@

That file doesn't exist and I think this patch doesn't want to patch it
anyway.

I'll just drop the vmlinux.lds.S chunk...

Re: [PATCH 1/22] UML - Include the linker script rather than symlink it

[Al Viro]

On Mon, May 02, 2005 at 05:06:54PM -0700, Andrew Morton wrote:
> That file doesn't exist and I think this patch doesn't want to patch it
> anyway.
> 
> I'll just drop the vmlinux.lds.S chunk...

Correct patch is on ftp.linux.org.uk/pub/people/viro/UM0-uml-ldscript-RC12-rc3

Short version of the story: current tree plays with a symlink from
vmlinux.lds.S to one of two files, depending on the config.  That doesn't
work well and confused vmlinux.lds.S chunk is actually a demonstration of
breakage - symlink had not been properly cleaned up.

New variant has _fixed_ vmlinux.lds.S - real file with
#include <linux/config.h>
#ifdef CONFIG_LD_SCRIPT_STATIC
#include "uml.lds.S"
#else
#include "dyn.lds.S"
#endif
in it and no symlink magic in makefiles.  Since we feed it through cpp
anyway, we can simply let cpp do all work - including picking the right
script depending on config.

Re: [PATCH 1/22] UML - Include the linker script rather than symlink it

[Andrew Morton]

Al Viro <viro@parcelfarce.linux.theplanet.co.uk> wrote:
>
> On Mon, May 02, 2005 at 05:06:54PM -0700, Andrew Morton wrote:
> > That file doesn't exist and I think this patch doesn't want to patch it
> > anyway.
> > 
> > I'll just drop the vmlinux.lds.S chunk...
> 
> Correct patch is on ftp.linux.org.uk/pub/people/viro/UM0-uml-ldscript-RC12-rc3

OK...

There's a bit of a tangle going on in arch/um/kernel/Makefile, but it's
fairly simple stuff.

I put a rolled-up patch against 2.6.12-rc3 at
http://www.zip.com.au/~akpm/linux/patches/stuff/x.bz2 if someone wants to
check it all.

Is this all considered post-2.6.12 material?

Re: [PATCH 1/22] UML - Include the linker script rather than symlink it

[Al Viro]

On Mon, May 02, 2005 at 05:44:05PM -0700, Andrew Morton wrote:
> 
> There's a bit of a tangle going on in arch/um/kernel/Makefile, but it's
> fairly simple stuff.
> 
> I put a rolled-up patch against 2.6.12-rc3 at
> http://www.zip.com.au/~akpm/linux/patches/stuff/x.bz2 if someone wants to
> check it all.

Broken, due to missing mk_sc patch (it should go before mk_thread one;
how the hell did the latter manage to apply at all?)

> Is this all considered post-2.6.12 material?

Once all patches are in there - up to Jeff ;-)  Seriously, kbuild patchkit
is decently tested and has obviously no impact on other architectures.  So
that one is up to maintainer of architecture in question...

Re: [PATCH 1/22] UML - Include the linker script rather than symlink it

[Andrew Morton]

Al Viro <viro@parcelfarce.linux.theplanet.co.uk> wrote:
>
> On Mon, May 02, 2005 at 05:44:05PM -0700, Andrew Morton wrote:
> > 
> > There's a bit of a tangle going on in arch/um/kernel/Makefile, but it's
> > fairly simple stuff.
> > 
> > I put a rolled-up patch against 2.6.12-rc3 at
> > http://www.zip.com.au/~akpm/linux/patches/stuff/x.bz2 if someone wants to
> > check it all.
> 
> Broken, due to missing mk_sc patch (it should go before mk_thread one;

OK.  I ended up with an odd-looking arch/um/sys-x86_64/util/Makefile:



# Copyright 2003 - 2004 Pathscale, Inc
# Released under the GPL

hostprogs-y	:= mk_sc mk_thread
always		:= $(hostprogs-y)

HOSTCFLAGS_mk_thread.o := -I$(objtree)/arch/um


Is mk_sc still supposed to be in there?

> how the hell did the latter manage to apply at all?)

I just "fixed" things.  I do it all the time.

> > Is this all considered post-2.6.12 material?
> 
> Once all patches are in there - up to Jeff ;-)  Seriously, kbuild patchkit
> is decently tested and has obviously no impact on other architectures.  So
> that one is up to maintainer of architecture in question...

OK..

Re: [PATCH 1/22] UML - Include the linker script rather than symlink it

[Al Viro]

On Mon, May 02, 2005 at 06:28:51PM -0700, Andrew Morton wrote:
> # Copyright 2003 - 2004 Pathscale, Inc
> # Released under the GPL
> 
> hostprogs-y	:= mk_sc mk_thread
> always		:= $(hostprogs-y)
> 
> HOSTCFLAGS_mk_thread.o := -I$(objtree)/arch/um
 
> Is mk_sc still supposed to be in there?

Yes, along with hostflags for it.

> > how the hell did the latter manage to apply at all?)
> 
> I just "fixed" things.  I do it all the time.

Erm...  "Gives rejects" => "ought to investigate" => "bugger it, let
submitter figure it out and resubmit" would be an obvious progression...

OK.  Please, pick the following patches:
UM0-uml-ldscript-RC12-rc3
UM1-uml-os-RC12-rc3
UM2-uml-user-constants-RC12-rc3
UM3-uml-ptregs-RC12-rc3
UM4-uml-sc-RC12-rc3
UM5-uml-kernel-offsets-RC12-rc3
UM6-uml-thread-RC12-rc3
UM7-uml-util-RC12-rc3
UM8-uml-clean-RC12-rc3
UM10-uml-O-RC12-rc3
in ftp.linux.org.uk/pub/people/viro.  In the order above.  That's where
the beginning of Jeff's series had come from (his 1--9 and 4.5).

(In case if you wonder about UM9, that one already has an equivalent in
-mm).

Re: [PATCH 1/22] UML - Include the linker script rather than symlink it

[Andrew Morton]

Al Viro <viro@parcelfarce.linux.theplanet.co.uk> wrote:
>
> On Mon, May 02, 2005 at 06:28:51PM -0700, Andrew Morton wrote:
> > # Copyright 2003 - 2004 Pathscale, Inc
> > # Released under the GPL
> > 
> > hostprogs-y	:= mk_sc mk_thread
> > always		:= $(hostprogs-y)
> > 
> > HOSTCFLAGS_mk_thread.o := -I$(objtree)/arch/um
>  
> > Is mk_sc still supposed to be in there?
> 
> Yes, along with hostflags for it.

Ah, OK.

> > > how the hell did the latter manage to apply at all?)
> > 
> > I just "fixed" things.  I do it all the time.
> 
> Erm...  "Gives rejects" => "ought to investigate" => "bugger it, let
> submitter figure it out and resubmit" would be an obvious progression...

Sometimes.  Most of the time it's less work (and much less latency) just to
fix stuff up.

> OK.  Please, pick the following patches:
> UM0-uml-ldscript-RC12-rc3
> UM1-uml-os-RC12-rc3
> UM2-uml-user-constants-RC12-rc3
> UM3-uml-ptregs-RC12-rc3
> UM4-uml-sc-RC12-rc3
> UM5-uml-kernel-offsets-RC12-rc3
> UM6-uml-thread-RC12-rc3
> UM7-uml-util-RC12-rc3
> UM8-uml-clean-RC12-rc3
> UM10-uml-O-RC12-rc3
> in ftp.linux.org.uk/pub/people/viro.  In the order above.  That's where
> the beginning of Jeff's series had come from (his 1--9 and 4.5).

OK, did that.  Had one reject against
uml-kbuild-avoid-useless-rebuilds.patch, but 'twas easily fixed.

Re: [PATCH 1/22] UML - Include the linker script rather than symlink it

[Jeff Dike]

On Mon, May 02, 2005 at 07:02:50PM -0700, Andrew Morton wrote:
> OK, did that.  Had one reject against
> uml-kbuild-avoid-useless-rebuilds.patch, but 'twas easily fixed.

Thanks.  Sorry about the hassle.  The vmlinux.lds botch was a symptom of
the problem Al was fixing (the uncleaned symlink in my tree causing a reject
in a clean tree), and the missing mk_sc patch was just an oversight on my
part.

				Jeff