Skip to contentSkip to conversation

livepatch: enable -f{function,data}-sections compiler option

PATCH Superseded Roger Pau Monne 4 participants 21 messages Mar 8, 2022 – Mar 9, 2022

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v3 cover livepatch: enable -f{function,data}-sections compiler option

Hello,

The content in v3 has been split in two patches, but is still mostly the same. The main difference is that first patch does a bit of cleanup of the build logic now that the header object file doesn't need to be the first one passed to the linker script.

Thanks, Roger.

Roger Pau Monne (2):
xen/build: put image header into a separate section
livepatch: set -f{function,data}-sections compiler option

 xen/Kconfig                 |  4 +++
 xen/Makefile                |  2 ++
 xen/arch/arm/arch.mk        |  2 --
 xen/arch/arm/arm32/Makefile |  3 +--
 xen/arch/arm/arm32/head.S   |  1 +
 xen/arch/arm/arm64/Makefile |  3 +--
 xen/arch/arm/arm64/head.S   |  1 +
 xen/arch/arm/xen.lds.S      | 49 ++++++++++++++++++++-----------------
 xen/arch/x86/Makefile       |  5 ++--
 xen/arch/x86/arch.mk        |  2 --
 xen/arch/x86/boot/head.S    |  2 +-
 xen/arch/x86/xen.lds.S      | 22 ++++++++++-------
 xen/common/Kconfig          |  1 +
 13 files changed, 54 insertions(+), 43 deletions(-)
Signature · 1 line
2.34.1
#1 lore raw

PATCH v3 1/2 xen/build: put image header into a separate section

So it can be explicitly placed ahead of the rest of the .text content in the linker script (and thus the resulting image). This is a prerequisite for further work that will add a catch-all to the text section (.text.*).

Note that placement of the sections inside of .text is also slightly adjusted to be more similar to the position found in the default GNU ld linker script.

The special handling of the object file containing the header data as the first object file passed to the linker command line can also be removed.

While there also remove the special handling of efi/ on x86. There's no need for the resulting object file to be passed in any special order to the linker.

---
Changes since v2:
- New in this version (split from patch 2).

---
 xen/arch/arm/arch.mk        |  2 --
 xen/arch/arm/arm32/Makefile |  3 +--
 xen/arch/arm/arm32/head.S   |  1 +
 xen/arch/arm/arm64/Makefile |  3 +--
 xen/arch/arm/arm64/head.S   |  1 +
 xen/arch/arm/xen.lds.S      |  8 ++++++--
 xen/arch/x86/Makefile       |  5 ++---
 xen/arch/x86/arch.mk        |  2 --
 xen/arch/x86/boot/head.S    |  2 +-
 xen/arch/x86/xen.lds.S      | 13 ++++++++-----
 10 files changed, 21 insertions(+), 19 deletions(-)
xen/arch/arm/arch.mk +0 -2
index 094b670723..58db76c4e1 100644
@@ -23,5 +23,3 @@ ifeq ($(CONFIG_ARM64_ERRATUM_843419),y)
         LDFLAGS += --fix-cortex-a53-843419
     endif
 endif
-
-ALL_OBJS-y := arch/arm/$(TARGET_SUBARCH)/head.o $(ALL_OBJS-y)
xen/arch/arm/arm32/Makefile +1 -2
index 3040eabce3..520fb42054 100644
@@ -4,11 +4,10 @@ obj-$(CONFIG_EARLY_PRINTK) += debug.o
 obj-y += domctl.o
 obj-y += domain.o
 obj-y += entry.o
+obj-y += head.o
 obj-y += insn.o
 obj-$(CONFIG_LIVEPATCH) += livepatch.o
 obj-y += proc-v7.o proc-caxx.o
 obj-y += smpboot.o
 obj-y += traps.o
 obj-y += vfp.o
-
-extra-y += head.o
xen/arch/arm/arm32/head.S +1 -0
index 7a906167ef..c837d3054c 100644
@@ -120,6 +120,7 @@
 
 #endif /* !CONFIG_EARLY_PRINTK */
 
+        .section .text.header, "ax", %progbits
         .arm
 
         /*
xen/arch/arm/arm64/Makefile +1 -2
index baa87655fa..6d507da0d4 100644
@@ -7,6 +7,7 @@ obj-$(CONFIG_EARLY_PRINTK) += debug.o
 obj-y += domctl.o
 obj-y += domain.o
 obj-y += entry.o
+obj-y += head.o
 obj-y += insn.o
 obj-$(CONFIG_LIVEPATCH) += livepatch.o
 obj-y += smc.o
@@ -14,5 +15,3 @@ obj-y += smpboot.o
 obj-y += traps.o
 obj-y += vfp.o
 obj-y += vsysreg.o
-
-extra-y += head.o
xen/arch/arm/arm64/head.S +1 -0
index 66d862fc81..e62c48ec1c 100644
@@ -133,6 +133,7 @@
         add \xb, \xb, x20
 .endm
 
+        .section .text.header, "ax", %progbits
         /*.aarch64*/
 
         /*
xen/arch/arm/xen.lds.S +6 -2
index 08016948ab..47d09d6cf1 100644
@@ -30,9 +30,13 @@ SECTIONS
   _start = .;
   .text : {
         _stext = .;            /* Text section */
+       *(.text.header)
+
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)
+
        *(.text)
-       *(.text.cold)
-       *(.text.unlikely)
+
        *(.fixup)
        *(.gnu.warning)
        _etext = .;             /* End of text section */
xen/arch/x86/Makefile +2 -3
index 9c40e0b4d7..04065a7310 100644
@@ -1,5 +1,7 @@
 obj-y += acpi/
+obj-y += boot/
 obj-y += cpu/
+obj-y += efi/
 obj-y += genapic/
 obj-$(CONFIG_GUEST) += guest/
 obj-$(CONFIG_HVM) += hvm/
@@ -77,9 +79,6 @@ obj-$(CONFIG_COMPAT) += x86_64/platform_hypercall.o
 obj-y += sysctl.o
 endif
 
-# Allows "clean" to descend into boot/
-subdir- += boot
-
 extra-y += asm-macros.i
 extra-y += xen.lds
 
xen/arch/x86/arch.mk +0 -2
index 8e57476d65..c90e56aeab 100644
@@ -117,5 +117,3 @@ endif
 
 # Set up the assembler include path properly for older toolchains.
 CFLAGS += -Wa,-I$(srctree)/include
-
-ALL_OBJS-y := arch/x86/boot/built_in.o arch/x86/efi/built_in.o $(ALL_OBJS-y)
xen/arch/x86/boot/head.S +1 -1
index dd1bea0d10..92d73345f0 100644
@@ -9,7 +9,7 @@
 #include <asm/cpufeature.h>
 #include <public/elfnote.h>
 
-        .text
+        .section .text.header, "ax", @progbits
         .code32
 
 #define sym_offs(sym)     ((sym) - __XEN_VIRT_START)
xen/arch/x86/xen.lds.S +8 -5
index 506bc8e404..715452aad9 100644
@@ -73,9 +73,7 @@ SECTIONS
   _start = .;
   DECL_SECTION(.text) {
         _stext = .;            /* Text and read-only data */
-       *(.text)
-       *(.text.__x86_indirect_thunk_*)
-       *(.text.page_aligned)
+       *(.text.header)
 
        . = ALIGN(PAGE_SIZE);
        _stextentry = .;
@@ -86,8 +84,13 @@ SECTIONS
        *(.text.kexec)          /* Page aligned in the object file. */
        kexec_reloc_end = .;
 
-       *(.text.cold)
-       *(.text.unlikely)
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)
+
+       *(.text)
+       *(.text.__x86_indirect_thunk_*)
+       *(.text.page_aligned)
+
        *(.fixup)
        *(.gnu.warning)
        _etext = .;             /* End of text section */
Signature · 1 line
2.34.1
#2 lore raw

Re: [PATCH v3 1/2] xen/build: put image header into a separate section

Quoted 13 lines · Roger Pau Monne

So it can be explicitly placed ahead of the rest of the .text content in the linker script (and thus the resulting image). This is a prerequisite for further work that will add a catch-all to the text section (.text.*).

Note that placement of the sections inside of .text is also slightly adjusted to be more similar to the position found in the default GNU ld linker script.

The special handling of the object file containing the header data as the first object file passed to the linker command line can also be removed.

Quoted 5 lines

While there also remove the special handling of efi/ on x86. There's no need for the resulting object file to be passed in any special order to the linker.

Looks good to me, but I have one question before feeling ready to offer R-b:

xen/arch/x86/xen.lds.S SECTIONS · line 88

Quoted 8 lines
xen/arch/x86/xen.lds.S +2 -2
@@ -86,8 +84,13 @@ SECTIONS
        *(.text.kexec)          /* Page aligned in the object file. */
        kexec_reloc_end = .;
 
-       *(.text.cold)
-       *(.text.unlikely)
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

What generates .text.*_unlikely? And if anything really does, why would .text.cold not have a similar equivalent?

Jan

#3 lore raw

Re: [PATCH v3 1/2] xen/build: put image header into a separate section

Quoted 27 lines · Jan Beulich
Quoted 13 lines · Roger Pau Monne

So it can be explicitly placed ahead of the rest of the .text content in the linker script (and thus the resulting image). This is a prerequisite for further work that will add a catch-all to the text section (.text.*).

Note that placement of the sections inside of .text is also slightly adjusted to be more similar to the position found in the default GNU ld linker script.

The special handling of the object file containing the header data as the first object file passed to the linker command line can also be removed.

Quoted 5 lines

While there also remove the special handling of efi/ on x86. There's no need for the resulting object file to be passed in any special order to the linker.

Looks good to me, but I have one question before feeling ready to offer R-b:

Quoted 4 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -86,8 +84,13 @@ SECTIONS
        *(.text.kexec)          /* Page aligned in the object file. */
        kexec_reloc_end = .;
 
Quoted 7 lines

xen/arch/x86/xen.lds.S SECTIONS · line 88

Quoted 4 lines
xen/arch/x86/xen.lds.S +2 -2
-       *(.text.cold)
-       *(.text.unlikely)
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

What generates .text.*_unlikely? And if anything really does, why would .text.cold not have a similar equivalent?

That matches what I saw in the default linker script from my version of GNU ld:

    *(.text.unlikely .text.*_unlikely .text.unlikely.*)

I really don't know what could generate .text.*_unlikely, but since it's part of the default linker script I assumed it was better to just add it.

Thanks, Roger.

Reviewed-by
#4 lore raw

Re: [PATCH v3 1/2] xen/build: put image header into a separate section

Quoted 36 lines · Roger Pau Monné
Quoted 27 lines · Jan Beulich
Quoted 13 lines · Roger Pau Monne

So it can be explicitly placed ahead of the rest of the .text content in the linker script (and thus the resulting image). This is a prerequisite for further work that will add a catch-all to the text section (.text.*).

Note that placement of the sections inside of .text is also slightly adjusted to be more similar to the position found in the default GNU ld linker script.

The special handling of the object file containing the header data as the first object file passed to the linker command line can also be removed.

Quoted 5 lines

While there also remove the special handling of efi/ on x86. There's no need for the resulting object file to be passed in any special order to the linker.

Looks good to me, but I have one question before feeling ready to offer R-b:

Quoted 4 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -86,8 +84,13 @@ SECTIONS
        *(.text.kexec)          /* Page aligned in the object file. */
        kexec_reloc_end = .;
 
Quoted 7 lines

xen/arch/x86/xen.lds.S SECTIONS · line 88

Quoted 4 lines
xen/arch/x86/xen.lds.S +2 -2
-       *(.text.cold)
-       *(.text.unlikely)
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

What generates .text.*_unlikely? And if anything really does, why would .text.cold not have a similar equivalent?

Quoted 8 lines

That matches what I saw in the default linker script from my version of GNU ld:

    *(.text.unlikely .text.*_unlikely .text.unlikely.*)

I really don't know what could generate .text.*_unlikely, but since it's part of the default linker script I assumed it was better to just add it.

I've checked - gcc up to 4.5.x would generate .text.*_unlikely; from 4.6.x. onwards it would be .text.unlikely.*.

As to the dissimilarity with .text.cold: I wonder why we have that in the first place. It matches our __cold attribute, just that we don't use that anywhere afaics.

In any event:

albeit preferably with .text.cold.* dropped again.

Jan

#5 lore raw

Re: [PATCH v3 1/2] xen/build: put image header into a separate section

Quoted 49 lines · Jan Beulich
Quoted 36 lines · Roger Pau Monné
Quoted 27 lines · Jan Beulich
Quoted 13 lines · Roger Pau Monne

So it can be explicitly placed ahead of the rest of the .text content in the linker script (and thus the resulting image). This is a prerequisite for further work that will add a catch-all to the text section (.text.*).

Note that placement of the sections inside of .text is also slightly adjusted to be more similar to the position found in the default GNU ld linker script.

The special handling of the object file containing the header data as the first object file passed to the linker command line can also be removed.

Quoted 5 lines

While there also remove the special handling of efi/ on x86. There's no need for the resulting object file to be passed in any special order to the linker.

Looks good to me, but I have one question before feeling ready to offer R-b:

Quoted 4 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -86,8 +84,13 @@ SECTIONS
        *(.text.kexec)          /* Page aligned in the object file. */
        kexec_reloc_end = .;
 
Quoted 7 lines

xen/arch/x86/xen.lds.S SECTIONS · line 88

Quoted 4 lines
xen/arch/x86/xen.lds.S +2 -2
-       *(.text.cold)
-       *(.text.unlikely)
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

What generates .text.*_unlikely? And if anything really does, why would .text.cold not have a similar equivalent?

Quoted 8 lines

That matches what I saw in the default linker script from my version of GNU ld:

    *(.text.unlikely .text.*_unlikely .text.unlikely.*)

I really don't know what could generate .text.*_unlikely, but since it's part of the default linker script I assumed it was better to just add it.

I've checked - gcc up to 4.5.x would generate .text.*_unlikely; from 4.6.x. onwards it would be .text.unlikely.*.

Quoted 7 lines

As to the dissimilarity with .text.cold: I wonder why we have that in the first place. It matches our __cold attribute, just that we don't use that anywhere afaics.

In any event:

albeit preferably with .text.cold.* dropped again.

Would you mind dropping the .text.cold.* at commit? Otherwise I can resend.

Thanks, Roger.

#6 lore raw

Re: [PATCH v3 1/2] xen/build: put image header into a separate section

Quoted 54 lines · Roger Pau Monné
Quoted 53 lines · Jan Beulich
Quoted 36 lines · Roger Pau Monné
Quoted 27 lines · Jan Beulich
Quoted 13 lines · Roger Pau Monne

So it can be explicitly placed ahead of the rest of the .text content in the linker script (and thus the resulting image). This is a prerequisite for further work that will add a catch-all to the text section (.text.*).

Note that placement of the sections inside of .text is also slightly adjusted to be more similar to the position found in the default GNU ld linker script.

The special handling of the object file containing the header data as the first object file passed to the linker command line can also be removed.

Quoted 5 lines

While there also remove the special handling of efi/ on x86. There's no need for the resulting object file to be passed in any special order to the linker.

Looks good to me, but I have one question before feeling ready to offer R-b:

Quoted 4 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -86,8 +84,13 @@ SECTIONS
        *(.text.kexec)          /* Page aligned in the object file. */
        kexec_reloc_end = .;
 
Quoted 7 lines

xen/arch/x86/xen.lds.S SECTIONS · line 88

Quoted 4 lines
xen/arch/x86/xen.lds.S +2 -2
-       *(.text.cold)
-       *(.text.unlikely)
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

What generates .text.*_unlikely? And if anything really does, why would .text.cold not have a similar equivalent?

Quoted 8 lines

That matches what I saw in the default linker script from my version of GNU ld:

    *(.text.unlikely .text.*_unlikely .text.unlikely.*)

I really don't know what could generate .text.*_unlikely, but since it's part of the default linker script I assumed it was better to just add it.

I've checked - gcc up to 4.5.x would generate .text.*_unlikely; from 4.6.x. onwards it would be .text.unlikely.*.

As to the dissimilarity with .text.cold: I wonder why we have that in the first place. It matches our __cold attribute, just that we don't use that anywhere afaics.

Quoted 6 lines
Quoted 3 lines

In any event:

albeit preferably with .text.cold.* dropped again.

Would you mind dropping the .text.cold.* at commit? Otherwise I can resend.

Sure; no need to resend just for this.

Jan

#7 lore raw

Re: [PATCH v3 1/2] xen/build: put image header into a separate section

In reply to Roger Pau Monne
Quoted 7 lines · Roger Pau Monne
xen/arch/arm/xen.lds.S +0 -0
index 08016948ab..47d09d6cf1 100644
@@ -30,9 +30,13 @@ SECTIONS
   _start = .;
   .text : {

xen/arch/arm/xen.lds.S SECTIONS · line 39

Quoted 10 lines
xen/arch/arm/xen.lds.S +6 -2
         _stext = .;            /* Text section */
+       *(.text.header)
+
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)
+
        *(.text)
-       *(.text.cold)
-       *(.text.unlikely)
+

Most of this hunk wants to move into patch 2.  Patch 1 should only introduce .text.header.

Same for x86.  Can fix on commit.

~Andrew

#8 lore raw

Re: [PATCH v3 1/2] xen/build: put image header into a separate section

Quoted 22 lines · Andrew Cooper
Quoted 7 lines · Roger Pau Monne
xen/arch/arm/xen.lds.S +0 -0
index 08016948ab..47d09d6cf1 100644
@@ -30,9 +30,13 @@ SECTIONS
   _start = .;
   .text : {

xen/arch/arm/xen.lds.S SECTIONS · line 39

Quoted 10 lines
xen/arch/arm/xen.lds.S +6 -2
         _stext = .;            /* Text section */
+       *(.text.header)
+
+       *(.text.cold .text.cold.*)
+       *(.text.unlikely .text.*_unlikely .text.unlikely.*)
+
        *(.text)
-       *(.text.cold)
-       *(.text.unlikely)
+

Most of this hunk wants to move into patch 2.  Patch 1 should only introduce .text.header.

Quoted 1 line

Same for x86.  Can fix on commit.

The justification for doing it here is to better match the ordering used by the default linker script from GNU ld. Obviously this also benefits the following patch. Feel free to move, but I think it's also fine to do it here.

Thanks, Roger.

#9 lore raw

PATCH v3 2/2 livepatch: set -f{function,data}-sections compiler option

In reply to Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

---
Changes since v2:
- Split the placing of the header code in a separate section to a pre-patch.
- Move Kconfig option to xen/Kconfig.
- Expand reasoning why .data.read_mostly needs to be moved on Arm.

Changes since v1:
- Introduce CC_SPLIT_SECTIONS for selecting the compiler options.
- Drop check for compiler options, all supported versions have them.
- Re-arrange section placement in .text, to match the default linker script.
- Introduce .text.header to contain the headers bits that must appear first in the final binary.
---
It seems on Arm the schedulers and hypfs .data sections should be moved into read_mostly.
---
Tested by gitlab in order to assert I didn't introduce any regression on Arm specially.

---
 xen/Kconfig            |  4 ++++
 xen/Makefile           |  2 ++
 xen/arch/arm/xen.lds.S | 41 +++++++++++++++++++++--------------------
 xen/arch/x86/xen.lds.S |  9 +++++----
 xen/common/Kconfig     |  1 +
 5 files changed, 33 insertions(+), 24 deletions(-)
xen/Kconfig +4 -0
index bcbd2758e5..d134397a0b 100644
@@ -27,6 +27,10 @@ config CLANG_VERSION
 config CC_HAS_VISIBILITY_ATTRIBUTE
 	def_bool $(cc-option,-fvisibility=hidden)
 
+# Use -f{function,data}-sections compiler parameters
+config CC_SPLIT_SECTIONS
+	bool
+
 source "arch/$(SRCARCH)/Kconfig"
 
 config DEFCONFIG_LIST
xen/Makefile +2 -0
index 5c21492d6f..18a4f7e101 100644
@@ -273,6 +273,8 @@ else
 CFLAGS += -fomit-frame-pointer
 endif
 
+CFLAGS-$(CONFIG_CC_SPLIT_SECTIONS) += -ffunction-sections -fdata-sections
+
 CFLAGS += -nostdinc -fno-builtin -fno-common
 CFLAGS += -Werror -Wredundant-decls -Wno-pointer-arith
 $(call cc-option-add,CFLAGS,CC,-Wvla)
xen/arch/arm/xen.lds.S +21 -20

70 lines not on this page · raw message

xen/arch/x86/xen.lds.S +5 -4
index 715452aad9..75925fe145 100644
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)
 
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)
 
   DECL_SECTION(.bss) {
@@ -309,7 +310,7 @@ SECTIONS
        *(.bss.percpu.read_mostly)
        . = ALIGN(SMP_CACHE_BYTES);
        __per_cpu_data_end = .;
-       *(.bss)
+       *(.bss .bss.*)
        . = ALIGN(POINTER_ALIGN);
        __bss_end = .;
   } PHDR(text)
xen/common/Kconfig +1 -0
index 6443943889..d921c74d61 100644
@@ -354,6 +354,7 @@ config LIVEPATCH
 	bool "Live patching support"
 	default X86
 	depends on "$(XEN_HAS_BUILD_ID)" = "y"
+	select CC_SPLIT_SECTIONS
 	---help---
 	  Allows a running Xen hypervisor to be dynamically patched using
 	  binary patches without rebooting. This is primarily used to binarily
Signature · 1 line
2.34.1
Reviewed-by on v3 2/2
#10 lore raw

Re: [PATCH v3 2/2] livepatch: set -f{function,data}-sections compiler option

Quoted 24 lines · Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

Quoted 7 lines

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

xen/arch/x86/xen.lds.S SECTIONS · line 95

Quoted 11 lines
xen/arch/x86/xen.lds.S +3 -0
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)

These last two now will not have any effect anymore when CC_SPLIT_SECTIONS=y. This may have undesirable effects on the overall size when there is more than one object with a .text.page_aligned contribution. In .data ...

xen/arch/x86/xen.lds.S SECTIONS · line 299

Quoted 9 lines
xen/arch/x86/xen.lds.S +1 -3
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)

... this continues to be named first. I wonder whether we wouldn't want to use SORT_BY_ALIGNMENT (if available) instead in both places.

Jan

#11 lore raw

Re: [PATCH v3 2/2] livepatch: set -f{function,data}-sections compiler option

Quoted 40 lines · Jan Beulich
Quoted 24 lines · Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

Quoted 7 lines

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

Quoted 5 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
Quoted 11 lines

xen/arch/x86/xen.lds.S SECTIONS · line 95

Quoted 6 lines
xen/arch/x86/xen.lds.S +3 -0
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)

These last two now will not have any effect anymore when CC_SPLIT_SECTIONS=y. This may have undesirable effects on the overall size when there is more than one object with a .text.page_aligned contribution. In .data ...

Agreed. I wondered whether to move those ahead of the main text section, so likely:

       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

       *(.text.page_aligned)
       *(.text.__x86_indirect_thunk_*)
       *(.text)
#ifdef CONFIG_CC_SPLIT_SECTIONS
       *(.text.*)

#endif

FWIW, Linux seems fine to package .text.page_aligned together with the rest of .text using the .text.[0-9a-zA-Z_]* catch-all.

Quoted 12 lines

xen/arch/x86/xen.lds.S SECTIONS · line 299

Quoted 9 lines
xen/arch/x86/xen.lds.S +1 -3
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)

... this continues to be named first. I wonder whether we wouldn't want to use SORT_BY_ALIGNMENT (if available) instead in both places.

We could use the command line option if available (--sort-section=alignment) to sort all wildcard sections?

Thanks, Roger.

#12 lore raw

Re: [PATCH v3 2/2] livepatch: set -f{function,data}-sections compiler option

Quoted 56 lines · Roger Pau Monné
Quoted 40 lines · Jan Beulich
Quoted 24 lines · Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

Quoted 7 lines

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

Quoted 5 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
Quoted 11 lines

xen/arch/x86/xen.lds.S SECTIONS · line 95

Quoted 6 lines
xen/arch/x86/xen.lds.S +3 -0
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)

These last two now will not have any effect anymore when CC_SPLIT_SECTIONS=y. This may have undesirable effects on the overall size when there is more than one object with a .text.page_aligned contribution. In .data ...

Agreed. I wondered whether to move those ahead of the main text section, so likely:

Quoted 8 lines
       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

       *(.text.page_aligned)
       *(.text.__x86_indirect_thunk_*)
       *(.text)
#ifdef CONFIG_CC_SPLIT_SECTIONS
       *(.text.*)

#endif

Perhaps; I'm not really worried of .text.__x86_indirect_thunk_*, though. When adding .text.* that one can likely go away.

Quoted 2 lines

FWIW, Linux seems fine to package .text.page_aligned together with the rest of .text using the .text.[0-9a-zA-Z_]* catch-all.

There's no question this is functionally fine. The question is how many extra padding areas are inserted because of this.

Quoted 13 lines
Quoted 12 lines

xen/arch/x86/xen.lds.S SECTIONS · line 299

Quoted 9 lines
xen/arch/x86/xen.lds.S +1 -3
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)

... this continues to be named first. I wonder whether we wouldn't want to use SORT_BY_ALIGNMENT (if available) instead in both places.

Quoted 2 lines

We could use the command line option if available (--sort-section=alignment) to sort all wildcard sections?

Depends on the scope of the sorting that would result when enabled globally like this.

Jan

#13 lore raw

Re: [PATCH v3 2/2] livepatch: set -f{function,data}-sections compiler option

Quoted 88 lines · Jan Beulich
Quoted 56 lines · Roger Pau Monné
Quoted 40 lines · Jan Beulich
Quoted 24 lines · Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

Quoted 7 lines

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

Quoted 5 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
Quoted 11 lines

xen/arch/x86/xen.lds.S SECTIONS · line 95

Quoted 6 lines
xen/arch/x86/xen.lds.S +3 -0
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)

These last two now will not have any effect anymore when CC_SPLIT_SECTIONS=y. This may have undesirable effects on the overall size when there is more than one object with a .text.page_aligned contribution. In .data ...

Agreed. I wondered whether to move those ahead of the main text section, so likely:

Quoted 8 lines
       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

       *(.text.page_aligned)
       *(.text.__x86_indirect_thunk_*)
       *(.text)
#ifdef CONFIG_CC_SPLIT_SECTIONS
       *(.text.*)

#endif

Perhaps; I'm not really worried of .text.__x86_indirect_thunk_*, though. When adding .text.* that one can likely go away.

Quoted 2 lines

FWIW, Linux seems fine to package .text.page_aligned together with the rest of .text using the .text.[0-9a-zA-Z_]* catch-all.

There's no question this is functionally fine. The question is how many extra padding areas are inserted because of this.

Quoted 13 lines
Quoted 12 lines

xen/arch/x86/xen.lds.S SECTIONS · line 299

Quoted 9 lines
xen/arch/x86/xen.lds.S +1 -3
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)

... this continues to be named first. I wonder whether we wouldn't want to use SORT_BY_ALIGNMENT (if available) instead in both places.

Quoted 5 lines
Quoted 2 lines

We could use the command line option if available (--sort-section=alignment) to sort all wildcard sections?

Depends on the scope of the sorting that would result when enabled globally like this.

I'm not sure I'm following. Don't we generally want to sort by alignment in order to avoid adding unnecessary padding as much as possible?

For any wildcard sections we really don't care anymore how they are sorted?

Thanks, Roger.

#14 lore raw

Re: [PATCH v3 2/2] livepatch: set -f{function,data}-sections compiler option

Quoted 95 lines · Roger Pau Monné
Quoted 88 lines · Jan Beulich
Quoted 56 lines · Roger Pau Monné
Quoted 40 lines · Jan Beulich
Quoted 24 lines · Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

Quoted 7 lines

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

Quoted 5 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
Quoted 11 lines

xen/arch/x86/xen.lds.S SECTIONS · line 95

Quoted 6 lines
xen/arch/x86/xen.lds.S +3 -0
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)

These last two now will not have any effect anymore when CC_SPLIT_SECTIONS=y. This may have undesirable effects on the overall size when there is more than one object with a .text.page_aligned contribution. In .data ...

Agreed. I wondered whether to move those ahead of the main text section, so likely:

Quoted 8 lines
       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

       *(.text.page_aligned)
       *(.text.__x86_indirect_thunk_*)
       *(.text)
#ifdef CONFIG_CC_SPLIT_SECTIONS
       *(.text.*)

#endif

Perhaps; I'm not really worried of .text.__x86_indirect_thunk_*, though. When adding .text.* that one can likely go away.

Quoted 2 lines

FWIW, Linux seems fine to package .text.page_aligned together with the rest of .text using the .text.[0-9a-zA-Z_]* catch-all.

There's no question this is functionally fine. The question is how many extra padding areas are inserted because of this.

Quoted 13 lines
Quoted 12 lines

xen/arch/x86/xen.lds.S SECTIONS · line 299

Quoted 9 lines
xen/arch/x86/xen.lds.S +1 -3
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)

... this continues to be named first. I wonder whether we wouldn't want to use SORT_BY_ALIGNMENT (if available) instead in both places.

Quoted 5 lines
Quoted 2 lines

We could use the command line option if available (--sort-section=alignment) to sort all wildcard sections?

Depends on the scope of the sorting that would result when enabled globally like this.

Quoted 6 lines

I'm not sure I'm following. Don't we generally want to sort by alignment in order to avoid adding unnecessary padding as much as possible?

For any wildcard sections we really don't care anymore how they are sorted?

Sure. Question is whether sorting is limited to within any single *(...) construct, or whether it could extend to adjacent ones. IOW whether the command line option strictly is a replacement of adding SORT_BY_ALIGNMENT to every one of these constructs.

Jan

#15 lore raw

Re: [PATCH v3 2/2] livepatch: set -f{function,data}-sections compiler option

Quoted 100 lines · Jan Beulich
Quoted 99 lines · Roger Pau Monné
Quoted 88 lines · Jan Beulich
Quoted 56 lines · Roger Pau Monné
Quoted 40 lines · Jan Beulich
Quoted 24 lines · Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

Quoted 7 lines

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

Quoted 5 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
Quoted 11 lines

xen/arch/x86/xen.lds.S SECTIONS · line 95

Quoted 6 lines
xen/arch/x86/xen.lds.S +3 -0
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)

These last two now will not have any effect anymore when CC_SPLIT_SECTIONS=y. This may have undesirable effects on the overall size when there is more than one object with a .text.page_aligned contribution. In .data ...

Agreed. I wondered whether to move those ahead of the main text section, so likely:

Quoted 8 lines
       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

       *(.text.page_aligned)
       *(.text.__x86_indirect_thunk_*)
       *(.text)
#ifdef CONFIG_CC_SPLIT_SECTIONS
       *(.text.*)

#endif

Perhaps; I'm not really worried of .text.__x86_indirect_thunk_*, though. When adding .text.* that one can likely go away.

Quoted 2 lines

FWIW, Linux seems fine to package .text.page_aligned together with the rest of .text using the .text.[0-9a-zA-Z_]* catch-all.

There's no question this is functionally fine. The question is how many extra padding areas are inserted because of this.

Quoted 13 lines
Quoted 12 lines

xen/arch/x86/xen.lds.S SECTIONS · line 299

Quoted 9 lines
xen/arch/x86/xen.lds.S +1 -3
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)

... this continues to be named first. I wonder whether we wouldn't want to use SORT_BY_ALIGNMENT (if available) instead in both places.

Quoted 5 lines
Quoted 2 lines

We could use the command line option if available (--sort-section=alignment) to sort all wildcard sections?

Depends on the scope of the sorting that would result when enabled globally like this.

I'm not sure I'm following. Don't we generally want to sort by alignment in order to avoid adding unnecessary padding as much as possible?

Quoted 7 lines
Quoted 2 lines

For any wildcard sections we really don't care anymore how they are sorted?

Sure. Question is whether sorting is limited to within any single *(...) construct, or whether it could extend to adjacent ones. IOW whether the command line option strictly is a replacement of adding SORT_BY_ALIGNMENT to every one of these constructs.

AFAICT the command line option will have the effect of setting the sorting of any wildcard containing sections to use SORT_BY_ALIGNMENT.
Ie: .data.* would become SORT_BY_ALIGNMENT(.data.*):

*(.data SORT_BY_ALIGNMENT(.data.*))

I've taken a look at the binutils ld source and that seems to be the case, any wildcard containing enum will get it's sorting set to by alignment (but I'm not familiar with ld code so I might be missing pieces).

Thanks, Roger.

#16 lore raw

Re: [PATCH v3 2/2] livepatch: set -f{function,data}-sections compiler option

Quoted 113 lines · Roger Pau Monné
Quoted 100 lines · Jan Beulich
Quoted 99 lines · Roger Pau Monné
Quoted 88 lines · Jan Beulich
Quoted 56 lines · Roger Pau Monné
Quoted 40 lines · Jan Beulich
Quoted 24 lines · Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

Quoted 7 lines

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

Quoted 5 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
Quoted 11 lines

xen/arch/x86/xen.lds.S SECTIONS · line 95

Quoted 6 lines
xen/arch/x86/xen.lds.S +3 -0
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)

These last two now will not have any effect anymore when CC_SPLIT_SECTIONS=y. This may have undesirable effects on the overall size when there is more than one object with a .text.page_aligned contribution. In .data ...

Agreed. I wondered whether to move those ahead of the main text section, so likely:

Quoted 8 lines
       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

       *(.text.page_aligned)
       *(.text.__x86_indirect_thunk_*)
       *(.text)
#ifdef CONFIG_CC_SPLIT_SECTIONS
       *(.text.*)

#endif

Perhaps; I'm not really worried of .text.__x86_indirect_thunk_*, though. When adding .text.* that one can likely go away.

Quoted 2 lines

FWIW, Linux seems fine to package .text.page_aligned together with the rest of .text using the .text.[0-9a-zA-Z_]* catch-all.

There's no question this is functionally fine. The question is how many extra padding areas are inserted because of this.

Quoted 13 lines
Quoted 12 lines

xen/arch/x86/xen.lds.S SECTIONS · line 299

Quoted 9 lines
xen/arch/x86/xen.lds.S +1 -3
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)

... this continues to be named first. I wonder whether we wouldn't want to use SORT_BY_ALIGNMENT (if available) instead in both places.

Quoted 5 lines
Quoted 2 lines

We could use the command line option if available (--sort-section=alignment) to sort all wildcard sections?

Depends on the scope of the sorting that would result when enabled globally like this.

I'm not sure I'm following. Don't we generally want to sort by alignment in order to avoid adding unnecessary padding as much as possible?

Quoted 7 lines
Quoted 2 lines

For any wildcard sections we really don't care anymore how they are sorted?

Sure. Question is whether sorting is limited to within any single *(...) construct, or whether it could extend to adjacent ones. IOW whether the command line option strictly is a replacement of adding SORT_BY_ALIGNMENT to every one of these constructs.

AFAICT the command line option will have the effect of setting the sorting of any wildcard containing sections to use SORT_BY_ALIGNMENT.
Ie: .data.* would become SORT_BY_ALIGNMENT(.data.*):

Quoted 6 lines

*(.data SORT_BY_ALIGNMENT(.data.*))

I've taken a look at the binutils ld source and that seems to be the case, any wildcard containing enum will get it's sorting set to by alignment (but I'm not familiar with ld code so I might be missing pieces).

Okay - why don't we try that then (in a separate patch, so it's going to be easy to revert)? For the patch here all I'd like to ask for is to keep .text.page_aligned enumerated explicitly (and the wildcard placed after it, obviously).

Jan

#17 lore raw

Re: [PATCH v3 2/2] livepatch: set -f{function,data}-sections compiler option

Quoted 105 lines · Jan Beulich
Quoted 104 lines · Roger Pau Monné
Quoted 91 lines · Jan Beulich
Quoted 90 lines · Roger Pau Monné
Quoted 81 lines · Jan Beulich
Quoted 53 lines · Roger Pau Monné
Quoted 39 lines · Jan Beulich
Quoted 24 lines · Roger Pau Monne

If livepatching support is enabled build the hypervisor with -f{function,data}-sections compiler options, which is required by the livepatching tools to detect changes and create livepatches.

This shouldn't result in any functional change on the hypervisor binary image, but does however require some changes in the linker script in order to handle that each function and data item will now be placed into its own section in object files. As a result add catch-all for .text, .data and .bss in order to merge each individual item section into the final image.

The main difference will be that .text.startup will end up being part of .text rather than .init, and thus won't be freed. .text.exit will also be part of .text rather than dropped. Overall this could make the image bigger, and package some .text code in a sub-optimal way.

On Arm the .data.read_mostly needs to be moved ahead of the .data section like it's already done on x86, so the .data.* catch-all doesn't also include .data.read_mostly. The alignment of .data.read_mostly also needs to be set to PAGE_SIZE so it doesn't end up being placed at the tail of a read-only page from the previous section. While there move the alignment of the .data section ahead of the section declaration, like it's done for other sections.

Quoted 7 lines

The benefit of having CONFIG_LIVEPATCH enable those compiler option is that the livepatch build tools no longer need to fiddle with the build system in order to enable them. Note the current livepatch tools are broken after the recent build changes due to the way they attempt to set -f{function,data}-sections.

Quoted 5 lines
xen/arch/x86/xen.lds.S +0 -0
@@ -88,6 +88,9 @@ SECTIONS
        *(.text.unlikely .text.*_unlikely .text.unlikely.*)
 
Quoted 10 lines

xen/arch/x86/xen.lds.S SECTIONS · line 95

Quoted 6 lines
xen/arch/x86/xen.lds.S +3 -0
        *(.text)
+#ifdef CONFIG_CC_SPLIT_SECTIONS
+       *(.text.*)
+#endif
        *(.text.__x86_indirect_thunk_*)
        *(.text.page_aligned)

These last two now will not have any effect anymore when CC_SPLIT_SECTIONS=y. This may have undesirable effects on the overall size when there is more than one object with a .text.page_aligned contribution. In .data ...

Agreed. I wondered whether to move those ahead of the main text section, so likely:

Quoted 8 lines
       *(.text.unlikely .text.*_unlikely .text.unlikely.*)

       *(.text.page_aligned)
       *(.text.__x86_indirect_thunk_*)
       *(.text)
#ifdef CONFIG_CC_SPLIT_SECTIONS
       *(.text.*)

#endif

Perhaps; I'm not really worried of .text.__x86_indirect_thunk_*, though. When adding .text.* that one can likely go away.

Quoted 2 lines

FWIW, Linux seems fine to package .text.page_aligned together with the rest of .text using the .text.[0-9a-zA-Z_]* catch-all.

There's no question this is functionally fine. The question is how many extra padding areas are inserted because of this.

Quoted 11 lines
Quoted 11 lines

xen/arch/x86/xen.lds.S SECTIONS · line 299

Quoted 9 lines
xen/arch/x86/xen.lds.S +1 -3
@@ -292,9 +295,7 @@ SECTIONS
 
   DECL_SECTION(.data) {
        *(.data.page_aligned)
-       *(.data)
-       *(.data.rel)
-       *(.data.rel.*)
+       *(.data .data.*)
   } PHDR(text)

... this continues to be named first. I wonder whether we wouldn't want to use SORT_BY_ALIGNMENT (if available) instead in both places.

Quoted 4 lines
Quoted 2 lines

We could use the command line option if available (--sort-section=alignment) to sort all wildcard sections?

Depends on the scope of the sorting that would result when enabled globally like this.

I'm not sure I'm following. Don't we generally want to sort by alignment in order to avoid adding unnecessary padding as much as possible?

Quoted 6 lines
Quoted 2 lines

For any wildcard sections we really don't care anymore how they are sorted?

Sure. Question is whether sorting is limited to within any single *(...) construct, or whether it could extend to adjacent ones. IOW whether the command line option strictly is a replacement of adding SORT_BY_ALIGNMENT to every one of these constructs.

AFAICT the command line option will have the effect of setting the sorting of any wildcard containing sections to use SORT_BY_ALIGNMENT.
Ie: .data.* would become SORT_BY_ALIGNMENT(.data.*):

*(.data SORT_BY_ALIGNMENT(.data.*))

Quoted 8 lines
Quoted 4 lines

I've taken a look at the binutils ld source and that seems to be the case, any wildcard containing enum will get it's sorting set to by alignment (but I'm not familiar with ld code so I might be missing pieces).

Okay - why don't we try that then (in a separate patch, so it's going to be easy to revert)? For the patch here all I'd like to ask for is to keep .text.page_aligned enumerated explicitly (and the wildcard placed after it, obviously).

I'm not convinced this will be an improvement.  It will make a marginal space saving, but cost runtime performance by breaking locality-of-reference inside a TU.

What would make an improvement (if this isn't how it already works) is having each TU sort by alignment on its own, then link in order.

~Andrew

Acked-by on the v3 series
#18 lore raw
In reply to Roger Pau Monne

Hi,

Quoted 7 lines · Roger Pau Monne

Hello,

The content in v3 has been split in two patches, but is still mostly the same. The main difference is that first patch does a bit of cleanup of the build logic now that the header object file doesn't need to be the first one passed to the linker script.

Quoted 5 lines

Thanks, Roger.

Roger Pau Monne (2):
xen/build: put image header into a separate section
livepatch: set -f{function,data}-sections compiler option

For the Arm bits:

Cheers,

Quoted 16 lines
  xen/Kconfig                 |  4 +++
  xen/Makefile                |  2 ++
  xen/arch/arm/arch.mk        |  2 --
  xen/arch/arm/arm32/Makefile |  3 +--
  xen/arch/arm/arm32/head.S   |  1 +
  xen/arch/arm/arm64/Makefile |  3 +--
  xen/arch/arm/arm64/head.S   |  1 +
  xen/arch/arm/xen.lds.S      | 49 ++++++++++++++++++++-----------------
  xen/arch/x86/Makefile       |  5 ++--
  xen/arch/x86/arch.mk        |  2 --
  xen/arch/x86/boot/head.S    |  2 +-
  xen/arch/x86/xen.lds.S      | 22 ++++++++++-------
  xen/common/Kconfig          |  1 +
  13 files changed, 54 insertions(+), 43 deletions(-)
Signature · 1 line
Julien Grall
#19 lore raw
Quoted 12 lines · Julien Grall

Hi,

Quoted 9 lines · Roger Pau Monne

Hello,

The content in v3 has been split in two patches, but is still mostly the same. The main difference is that first patch does a bit of cleanup of the build logic now that the header object file doesn't need to be the first one passed to the linker script.

Thanks, Roger.

Quoted 7 lines
Quoted 3 lines

Roger Pau Monne (2):
xen/build: put image header into a separate section
livepatch: set -f{function,data}-sections compiler option

For the Arm bits:

Thanks!

I don't think Acks require a scope tag, as you can only Ack changes to the code you are a maintainer for?

OTOH a 'Reviewed-by:' tag would indeed need to be limited if it only means to cover xen/arm bits.

In any case I've added the tag as provided to v4.

Roger.

#20 lore raw

Hi Roger,

Quoted 17 lines · Roger Pau Monné
Quoted 16 lines · Julien Grall

Hi,

Quoted 12 lines · Roger Pau Monne

Hello,

The content in v3 has been split in two patches, but is still mostly the same. The main difference is that first patch does a bit of cleanup of the build logic now that the header object file doesn't need to be the first one passed to the linker script.

Thanks, Roger.

Roger Pau Monne (2):
xen/build: put image header into a separate section
livepatch: set -f{function,data}-sections compiler option

Quoted 8 lines
Quoted 3 lines

For the Arm bits:

Thanks!

I don't think Acks require a scope tag, as you can only Ack changes to the code you are a maintainer for?

It is not required (same for reviewed-by). But it can be used to specify exactly which part you ack.

In this case, I am a maintainer for both common and Arm. I chose to ack Arm and leave someone else to confirm the change in common is fine.

Furthermore, with my committer hat on, I find handy to know which person acked what. This makes easier to check that the code is suitably acked.

Cheers,

[1] https://xenbits.xen.org/gitweb/?p=xen.git;a=blob;f=docs/process/sending-patches.pandoc;h=7ff7826c992b68804b41cc4a8605329f7a10e44c;hb=refs/heads/staging#l241

Signature · 1 line
Julien Grall