The current method to include 32 bit C boot code is:
- compile each function we want to use into a separate object file;
- each function is compiled with -fpic option;
- convert these object files to binary files. This operation removes GOP
which we don't want in the executable;
- a small assembly part in each file add the entry point;
- code can't have external references, all possible variables are passed
by value or pointer;
- include these binary files in head.S.
There are currently some limitations:
- code is compiled separately, it's not possible to share a function
(like memcpy) between different functions to use;
- although code is compiled with -fpic there's no certainty there are
no relocations, specifically data ones. This can lead into hard to
find bugs;
- it's hard to add a simple function;
- having to pass external variables makes hard to do multiple things
otherwise functions would require a lot of parameters so code would
have to be split into multiple functions which is not easy;
- we generate a single text section containing data and code, not a
problem at the moment but if we want to add W^X protection it's
not helpful.
Current change extends the current process:
- all object files are linked together before getting converted making
possible to share code between the function we want to call;
- a single object file is generated with all functions to use and
exported symbols to easily call;
- variables to use are declared in linker script and easily used inside
C code. Declaring them manually could be annoying but makes also
easier to check them. Using external pointers can be still an issue if
they are not fixed. If an external symbol is not declared this gives a
link error;
- linker script put data (bss and data) into a separate section and
check that that section is empty making sure code is W^X compatible;
Some details of the implementation:
- C code is compiled with -fpic flags (as before);
- object files from C code are linked together;
- the single bundled object file is linked with 2 slightly different
script files to generate 2 bundled object files;
- the 2 bundled object files are converted to binary removing the need
for global offset tables;
- a Python script is used to generate assembly source from the 2
binaries;
- the single assembly file is compiled to generate final bundled object
file;
- to detect possible unwanted relocation in data/code code is generated
with different addresses. This is enforced starting .text section at
different positions and adding a fixed "gap" at the beginning.
This makes sure code and data is position independent;
- to detect used symbols in data/code symbols are placed in .text
section at different offsets (based on the line in the linker script).
This is needed as potentially a reference to a symbol is converted to
a reference to the containing section so multiple symbols could be
converted to reference to same symbol (section name) and we need to
distinguish them;
- to avoid relocations
- --orphan-handling=error option to linker is used to make sure we
account for all possible sections from C code;
Current limitations:
- the main one is the lack of support for 64 bit code. It would make
sure that even the code used for 64 bit (at the moment EFI code) is
code and data position independent. We cannot assume that code that
came from code compiled for 32 bit and compiled for 64 bit is code and
data position independent, different compiler options lead to
different code/data.
Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com>
---
Changes since v1:
- separate lines adding files in Makefile;
- remove unneeded "#undef ENTRY" in build32.lds.S;
- print some information from combine_two_binaries only passing --verbose;
- detect --orphan-handling=error option dynamically;
- define and use a LD32;
- more intermediate targets to build more in parallel;
- use obj32 in Makefile to keep list of 32 bit object files;
- 32 bit object files are now named XXX.32.o;
- rename "cbundle" to "built_in_32".
---
xen/arch/x86/boot/.gitignore | 5 +-
xen/arch/x86/boot/Makefile | 58 +++--
.../x86/boot/{build32.lds => build32.lds.S} | 43 +++-
xen/arch/x86/boot/cmdline.c | 12 --
xen/arch/x86/boot/head.S | 12 --
xen/arch/x86/boot/reloc.c | 14 --
xen/tools/combine_two_binaries | 198 ++++++++++++++++++
7 files changed, 283 insertions(+), 59 deletions(-)
rename xen/arch/x86/boot/{build32.lds => build32.lds.S} (63%)
create mode 100755 xen/tools/combine_two_binaries
diff --git a/xen/arch/x86/boot/.gitignore b/xen/arch/x86/boot/.gitignore
index a379db7988..ebad650e5c 100644
--- a/xen/arch/x86/boot/.gitignore
+++ b/xen/arch/x86/boot/.gitignore
@@ -1,3 +1,4 @@
/mkelf32
-/*.bin
-/*.lnk
+/build32.*.lds
+/built_in_32.*.bin
+/built_in_32.*.map
diff --git a/xen/arch/x86/boot/Makefile b/xen/arch/x86/boot/Makefile
index ff0f965876..4cf0d7e140 100644
--- a/xen/arch/x86/boot/Makefile
+++ b/xen/arch/x86/boot/Makefile
@@ -1,15 +1,18 @@
+obj32 := cmdline.o
+obj32 += reloc.o
+
obj-bin-y += head.o
+obj-bin-y += built_in_32.o
-head-bin-objs := cmdline.o reloc.o
+obj32 := $(patsubst %.o,%.32.o,$(obj32))
-nocov-y += $(head-bin-objs)
-noubsan-y += $(head-bin-objs)
-targets += $(head-bin-objs)
+nocov-y += $(obj32)
+noubsan-y += $(obj32)
+targets += $(obj32)
-head-bin-objs := $(addprefix $(obj)/,$(head-bin-objs))
+obj32 := $(addprefix $(obj)/,$(obj32))
$(obj)/head.o: AFLAGS-y += -Wa$(comma)-I$(obj)
-$(obj)/head.o: $(head-bin-objs:.o=.bin)
CFLAGS_x86_32 := $(subst -m64,-m32 -march=i686,$(XEN_TREEWIDE_CFLAGS))
$(call cc-options-add,CFLAGS_x86_32,CC,$(EMBEDDED_EXTRA_CFLAGS))
@@ -17,17 +20,46 @@ CFLAGS_x86_32 += -Werror -fno-builtin -g0 -msoft-float -mregparm=3
CFLAGS_x86_32 += -nostdinc -include $(filter %/include/xen/config.h,$(XEN_CFLAGS))
CFLAGS_x86_32 += $(filter -I% -O%,$(XEN_CFLAGS)) -D__XEN__
+LD32 := $(LD) $(subst x86_64,i386,$(LDFLAGS_DIRECT))
+
# override for 32bit binaries
-$(head-bin-objs): CFLAGS_stack_boundary :=
-$(head-bin-objs): XEN_CFLAGS := $(CFLAGS_x86_32) -fpic
+$(obj32): CFLAGS_stack_boundary :=
+$(obj32): XEN_CFLAGS := $(CFLAGS_x86_32) -fpic
LDFLAGS_DIRECT-$(call ld-option,--warn-rwx-segments) := --no-warn-rwx-segments
LDFLAGS_DIRECT += $(LDFLAGS_DIRECT-y)
-%.bin: %.lnk
- $(OBJCOPY) -j .text -O binary $< $@
+$(obj)/%.32.o: $(src)/%.c FORCE
+ $(call if_changed_rule,cc_o_c)
+
+$(obj)/build32.final.lds: AFLAGS-y += -DFINAL
+$(obj)/build32.other.lds $(obj)/build32.final.lds: $(src)/build32.lds.S
+ $(call if_changed_dep,cpp_lds_S)
+
+orphan-handling-$(call ld-option,--orphan-handling=error) := --orphan-handling=error
+
+# link all object files together
+$(obj)/built_in_32.tmp.o: $(obj32)
+ $(LD32) -r -o $@ $(obj32)
+
+$(obj)/built_in_32.%.bin: $(obj)/build32.%.lds $(obj)/built_in_32.tmp.o
+## link bundle with a given layout
+ $(LD32) $(orphan-handling-y) -N -T $< -Map $(obj)/built_in_32.$(*F).map -o $(obj)/built_in_32.$(*F).o $(obj)/built_in_32.tmp.o
+## extract binaries from object
+ $(OBJCOPY) -j .text -O binary $(obj)/built_in_32.$(*F).o $@
+ rm -f $(obj)/built_in_32.$(*F).o
-%.lnk: %.o $(src)/build32.lds
- $(LD) $(subst x86_64,i386,$(LDFLAGS_DIRECT)) -N -T $(filter %.lds,$^) -o $@ $<
+# generate final object file combining and checking above binaries
+$(obj)/built_in_32.o: $(obj)/built_in_32.other.bin $(obj)/built_in_32.final.bin
+ $(PYTHON) $(srctree)/tools/combine_two_binaries \
+ --script $(obj)/build32.final.lds \
+ --bin1 $(obj)/built_in_32.other.bin --bin2 $(obj)/built_in_32.final.bin \
+ --map $(obj)/built_in_32.final.map \
+ --exports cmdline_parse_early,reloc \
+ --section-header '.section .init.text, "ax", @progbits' \
+ --output $(obj)/built_in_32.s
+ $(CC) -c $(obj)/built_in_32.s -o $@.tmp
+ rm -f $(obj)/built_in_32.s $@
+ mv $@.tmp $@
-clean-files := *.lnk *.bin
+clean-files := built_in_32.*.bin built_in_32.*.map build32.*.lds
diff --git a/xen/arch/x86/boot/build32.lds b/xen/arch/x86/boot/build32.lds.S
similarity index 63%
rename from xen/arch/x86/boot/build32.lds
rename to xen/arch/x86/boot/build32.lds.S
index 56edaa727b..72a4c5c614 100644
--- a/xen/arch/x86/boot/build32.lds
+++ b/xen/arch/x86/boot/build32.lds.S
@@ -15,22 +15,52 @@
* with this program. If not, see <http://www.gnu.org/licenses/>.
*/
-ENTRY(_start)
+#ifdef FINAL
+# define GAP 0
+# define MULT 0
+# define TEXT_START
+#else
+# define GAP 0x010200
+# define MULT 1
+# define TEXT_START 0x408020
+#endif
+# define DECLARE_IMPORT(name) name = . + (__LINE__ * MULT)
+
+ENTRY(dummy_start)
SECTIONS
{
- /* Merge code and data into one section. */
- .text : {
+ /* Merge code and read-only data into one section. */
+ .text TEXT_START : {
+ /* Silence linker warning, we are not going to use it */
+ dummy_start = .;
+
+ /* Declare below any symbol name needed.
+ * Each symbol should be on its own line.
+ * It looks like a tedious work but we make sure the things we use.
+ * Potentially they should be all variables. */
+ DECLARE_IMPORT(__base_relocs_start);
+ DECLARE_IMPORT(__base_relocs_end);
+ . = . + GAP;
*(.text)
*(.text.*)
- *(.data)
- *(.data.*)
*(.rodata)
*(.rodata.*)
+ }
+
+ /* Writeable data sections. Check empty.
+ * We collapse all into code section and we don't want it to be writeable. */
+ .data : {
+ *(.data)
+ *(.data.*)
*(.bss)
*(.bss.*)
}
-
+ /DISCARD/ : {
+ *(.comment)
+ *(.comment.*)
+ *(.note.*)
+ }
/* Dynamic linkage sections. Collected simply so we can check they're empty. */
.got : {
*(.got)
@@ -64,3 +94,4 @@ ASSERT(SIZEOF(.igot.plt) == 0, ".igot.plt non-empty")
ASSERT(SIZEOF(.iplt) == 0, ".iplt non-empty")
ASSERT(SIZEOF(.plt) == 0, ".plt non-empty")
ASSERT(SIZEOF(.rel) == 0, "leftover relocations")
+ASSERT(SIZEOF(.data) == 0, "we don't want data")
diff --git a/xen/arch/x86/boot/cmdline.c b/xen/arch/x86/boot/cmdline.c
index fc9241ede9..196c580e91 100644
--- a/xen/arch/x86/boot/cmdline.c
+++ b/xen/arch/x86/boot/cmdline.c
@@ -18,18 +18,6 @@
* Linux kernel source (linux/lib/string.c).
*/
-/*
- * This entry point is entered from xen/arch/x86/boot/head.S with:
- * - %eax = &cmdline,
- * - %edx = &early_boot_opts.
- */
-asm (
- " .text \n"
- " .globl _start \n"
- "_start: \n"
- " jmp cmdline_parse_early \n"
- );
-
#include <xen/compiler.h>
#include <xen/kconfig.h>
#include <xen/macros.h>
diff --git a/xen/arch/x86/boot/head.S b/xen/arch/x86/boot/head.S
index c4de1dfab5..e0776e3896 100644
--- a/xen/arch/x86/boot/head.S
+++ b/xen/arch/x86/boot/head.S
@@ -759,18 +759,6 @@ trampoline_setup:
/* Jump into the relocated trampoline. */
lret
- /*
- * cmdline and reloc are written in C, and linked to be 32bit PIC with
- * entrypoints at 0 and using the fastcall convention.
- */
-FUNC_LOCAL(cmdline_parse_early)
- .incbin "cmdline.bin"
-END(cmdline_parse_early)
-
-FUNC_LOCAL(reloc)
- .incbin "reloc.bin"
-END(reloc)
-
ENTRY(trampoline_start)
#include "trampoline.S"
ENTRY(trampoline_end)
diff --git a/xen/arch/x86/boot/reloc.c b/xen/arch/x86/boot/reloc.c
index 8c58affcd9..94b078d7b1 100644
--- a/xen/arch/x86/boot/reloc.c
+++ b/xen/arch/x86/boot/reloc.c
@@ -12,20 +12,6 @@
* Daniel Kiper <daniel.kiper@oracle.com>
*/
-/*
- * This entry point is entered from xen/arch/x86/boot/head.S with:
- * - %eax = MAGIC,
- * - %edx = INFORMATION_ADDRESS,
- * - %ecx = TOPMOST_LOW_MEMORY_STACK_ADDRESS.
- * - 0x04(%esp) = BOOT_VIDEO_INFO_ADDRESS.
- */
-asm (
- " .text \n"
- " .globl _start \n"
- "_start: \n"
- " jmp reloc \n"
- );
-
#include <xen/compiler.h>
#include <xen/macros.h>
#include <xen/types.h>
diff --git a/xen/tools/combine_two_binaries b/xen/tools/combine_two_binaries
new file mode 100755
index 0000000000..ea2d6ddc4e
--- /dev/null
+++ b/xen/tools/combine_two_binaries
@@ -0,0 +1,198 @@
+#!/usr/bin/env python3
+
+from __future__ import print_function
+import argparse
+import re
+import struct
+import sys
+
+parser = argparse.ArgumentParser(description='Generate assembly file to merge into other code.')
+parser.add_argument('--script', dest='script',
+ required=True,
+ help='Linker script for extracting symbols')
+parser.add_argument('--bin1', dest='bin1',
+ required=True,
+ help='First binary')
+parser.add_argument('--bin2', dest='bin2',
+ required=True,
+ help='Second binary')
+parser.add_argument('--output', dest='output',
+ help='Output file')
+parser.add_argument('--map', dest='mapfile',
+ help='Map file to read for symbols to export')
+parser.add_argument('--exports', dest='exports',
+ help='Symbols to export')
+parser.add_argument('--section-header', dest='section_header',
+ default='.text',
+ help='Section header declaration')
+parser.add_argument('-v', '--verbose',
+ action='store_true')
+args = parser.parse_args()
+
+gap = 0x010200
+text_diff = 0x408020
+
+# Parse linker script for external symbols to use.
+symbol_re = re.compile(r'\s+(\S+)\s*=\s*\.\s*\+\s*\((\d+)\s*\*\s*0\s*\)\s*;')
+symbols = {}
+lines = {}
+for line in open(args.script):
+ m = symbol_re.match(line)
+ if not m:
+ continue
+ (name, line_num) = (m.group(1), int(m.group(2)))
+ if line_num == 0:
+ raise Exception("Invalid line number found:\n\t" + line)
+ if line_num in symbols:
+ raise Exception("Symbol with this line already present:\n\t" + line)
+ if name in lines:
+ raise Exception("Symbol with this name already present:\n\t" + name)
+ symbols[line_num] = name
+ lines[name] = line_num
+
+exports = []
+if args.exports is not None:
+ exports = dict([(name, None) for name in args.exports.split(',')])
+
+# Parse mapfile, look for ther symbols we want to export.
+if args.mapfile is not None:
+ symbol_re = re.compile(r'\s{15,}0x([0-9a-f]+)\s+(\S+)\n')
+ for line in open(args.mapfile):
+ m = symbol_re.match(line)
+ if not m or m.group(2) not in exports:
+ continue
+ addr = int(m.group(1), 16)
+ exports[m.group(2)] = addr
+for (name, addr) in exports.items():
+ if addr is None:
+ raise Exception("Required export symbols %s not found" % name)
+
+file1 = open(args.bin1, 'rb')
+file2 = open(args.bin2, 'rb')
+file1.seek(0, 2)
+size1 = file1.tell()
+file2.seek(0, 2)
+size2 = file2.tell()
+if size1 > size2:
+ file1, file2 = file2, file1
+ size1, size2 = size2, size1
+if size2 != size1 + gap:
+ raise Exception('File sizes do not match')
+
+file1.seek(0, 0)
+data1 = file1.read(size1)
+file2.seek(gap, 0)
+data2 = file2.read(size1)
+
+max_line = max(symbols.keys())
+
+def to_int32(n):
+ '''Convert a number to signed 32 bit integer truncating if needed'''
+ mask = (1 << 32) - 1
+ h = 1 << 31
+ return (n & mask) ^ h - h
+
+i = 0
+references = {}
+internals = 0
+while i <= size1 - 4:
+ n1 = struct.unpack('<I', data1[i:i+4])[0]
+ n2 = struct.unpack('<I', data2[i:i+4])[0]
+ i += 1
+ # The two numbers are the same, no problems
+ if n1 == n2:
+ continue
+ # Try to understand why they are different
+ diff = to_int32(n1 - n2)
+ if diff == -gap: # this is an internal relocation
+ pos = i - 1
+ print(("Internal relocation found at position %#x "
+ "n1=%#x n2=%#x diff=%#x") % (pos, n1, n2, diff),
+ file=sys.stderr)
+ i += 3
+ internals += 1
+ if internals >= 10:
+ break
+ continue
+ # This is a relative relocation to a symbol, accepted, code/data is
+ # relocatable.
+ if diff < gap and diff >= gap - max_line:
+ n = gap - diff
+ symbol = symbols.get(n)
+ # check we have a symbol
+ if symbol is None:
+ raise Exception("Cannot find symbol for line %d" % n)
+ pos = i - 1
+ if args.verbose:
+ print('Position %#x %d %s' % (pos, n, symbol), file=sys.stderr)
+ i += 3
+ references[pos] = symbol
+ continue
+ # First byte is the same, move to next byte
+ if diff & 0xff == 0 and i <= size1 - 4:
+ continue
+ # Probably a type of relocation we don't want or support
+ pos = i - 1
+ suggestion = ''
+ symbol = symbols.get(-diff - text_diff)
+ if symbol is not None:
+ suggestion = " Maybe %s is not defined as hidden?" % symbol
+ raise Exception(("Unexpected difference found at %#x "
+ "n1=%#x n2=%#x diff=%#x gap=%#x.%s") % \
+ (pos, n1, n2, diff, gap, suggestion))
+if internals != 0:
+ raise Exception("Previous relocations found")
+
+def line_bytes(buf, out):
+ '''Output an assembly line with all bytes in "buf"'''
+ if type(buf) == str:
+ print("\t.byte " + ','.join([str(ord(c)) for c in buf]), file=out)
+ else:
+ print("\t.byte " + ','.join([str(n) for n in buf]), file=out)
+
+def part(start, end, out):
+ '''Output bytes of "data" from "start" to "end"'''
+ while start < end:
+ e = min(start + 16, end)
+ line_bytes(data1[start:e], out)
+ start = e
+
+def reference(pos, out):
+ name = references[pos]
+ n = struct.unpack('<I', data1[pos:pos+4])[0]
+ sign = '+'
+ if n >= (1 << 31):
+ n -= (1 << 32)
+ n += pos
+ if n < 0:
+ n = -n
+ sign = '-'
+ print("\t.hidden %s\n\t.long %s %s %#x - ." % (name, name, sign, n),
+ file=out)
+
+def output(out):
+ prev = 0
+ exports_by_addr = {}
+ for (sym, addr) in exports.items():
+ exports_by_addr.setdefault(addr, []).append(sym)
+ positions = list(references.keys())
+ positions += list(exports_by_addr.keys())
+ for pos in sorted(positions):
+ part(prev, pos, out)
+ prev = pos
+ if pos in references:
+ reference(pos, out)
+ prev = pos + 4
+ if pos in exports_by_addr:
+ for sym in exports_by_addr[pos]:
+ print("\t.global %s\n\t.hidden %s\n%s:" % (sym, sym, sym),
+ file=out)
+ part(prev, size1, out)
+
+out = sys.stdout
+if args.output is not None:
+ out = open(args.output, 'w')
+print('\t' + args.section_header, file=out)
+output(out)
+print('\n\t.section\t.note.GNU-stack,"",@progbits', file=out)
+out.flush()
--
2.34.1On 11.10.2024 10:52, Frediano Ziglio wrote: > diff --git a/xen/tools/combine_two_binaries b/xen/tools/combine_two_binaries > new file mode 100755 > index 0000000000..ea2d6ddc4e > --- /dev/null > +++ b/xen/tools/combine_two_binaries > @@ -0,0 +1,198 @@ > +#!/usr/bin/env python3 Nit: Such files, by convention, are named *.py, I think. Jan
On 11/10/2024 2:20 pm, Jan Beulich wrote: > On 11.10.2024 10:52, Frediano Ziglio wrote: >> diff --git a/xen/tools/combine_two_binaries b/xen/tools/combine_two_binaries >> new file mode 100755 >> index 0000000000..ea2d6ddc4e >> --- /dev/null >> +++ b/xen/tools/combine_two_binaries >> @@ -0,0 +1,198 @@ >> +#!/usr/bin/env python3 > Nit: Such files, by convention, are named *.py, I think. Yes please, and I forgot it on my python review. ~Andrew
On 11/10/2024 9:52 am, Frediano Ziglio wrote:
> diff --git a/xen/arch/x86/boot/Makefile b/xen/arch/x86/boot/Makefile
> index ff0f965876..4cf0d7e140 100644
> --- a/xen/arch/x86/boot/Makefile
> +++ b/xen/arch/x86/boot/Makefile
> ...
> +$(obj)/built_in_32.o: $(obj)/built_in_32.other.bin $(obj)/built_in_32.final.bin
> + $(PYTHON) $(srctree)/tools/combine_two_binaries \
> + --script $(obj)/build32.final.lds \
> + --bin1 $(obj)/built_in_32.other.bin --bin2 $(obj)/built_in_32.final.bin \
> + --map $(obj)/built_in_32.final.map \
> + --exports cmdline_parse_early,reloc \
> + --section-header '.section .init.text, "ax", @progbits' \
> + --output $(obj)/built_in_32.s
I can't see a case where we'd want this anywhere other than .init.text,
so I'd drop the --section-header and just write it out unconditionally.
However, looking at the output:
xen$ head arch/x86/boot/built_in_32.S
.section .init.text, "ax", @progbits
.byte 137,194,128,56,0,116,6,64,128,56,0,117,250,41,208,195
.byte 133,201,116,42,86,83,141,52,8,64,15,182,72,255,66,15
.byte 182,90,255,56,217,117,15,132,201,116,25,57,198,117,234,184
This wants to start with a comment saying that it was automatically
generated by combine_two_binaries.
Next, we need some kind of symbol at the start. Right now, the
disassembly reads:
arch/x86/boot/built_in_32.o: file format elf64-x86-64
Disassembly of section .init.text:
0000000000000000 <cmdline_parse_early-0x391>:
0: 89 c2 mov %eax,%edx
2: 80 38 00 cmpb $0x0,(%eax)
because most metadata is lost by this transformation, and it doesn't
know that this is in fact strlen(). I'd suggest suggest obj32_start: or
equivalent.
> diff --git a/xen/arch/x86/boot/build32.lds b/xen/arch/x86/boot/build32.lds.S
> similarity index 63%
> rename from xen/arch/x86/boot/build32.lds
> rename to xen/arch/x86/boot/build32.lds.S
> index 56edaa727b..72a4c5c614 100644
> --- a/xen/arch/x86/boot/build32.lds
> +++ b/xen/arch/x86/boot/build32.lds.S
> @@ -15,22 +15,52 @@
> * with this program. If not, see <http://www.gnu.org/licenses/>.
> */
>
> -ENTRY(_start)
> +#ifdef FINAL
> +# define GAP 0
> +# define MULT 0
> +# define TEXT_START
> +#else
> +# define GAP 0x010200
> +# define MULT 1
> +# define TEXT_START 0x408020
> +#endif
> +# define DECLARE_IMPORT(name) name = . + (__LINE__ * MULT)
> +
> +ENTRY(dummy_start)
>
> SECTIONS
> {
> - /* Merge code and data into one section. */
> - .text : {
> + /* Merge code and read-only data into one section. */
> + .text TEXT_START : {
> + /* Silence linker warning, we are not going to use it */
> + dummy_start = .;
> +
> + /* Declare below any symbol name needed.
> + * Each symbol should be on its own line.
> + * It looks like a tedious work but we make sure the things we use.
> + * Potentially they should be all variables. */
> + DECLARE_IMPORT(__base_relocs_start);
> + DECLARE_IMPORT(__base_relocs_end);
> + . = . + GAP;
> *(.text)
> *(.text.*)
> - *(.data)
> - *(.data.*)
> *(.rodata)
> *(.rodata.*)
> + }
> +
> + /* Writeable data sections. Check empty.
> + * We collapse all into code section and we don't want it to be writeable. */
> + .data : {
> + *(.data)
> + *(.data.*)
> *(.bss)
> *(.bss.*)
> }
> -
> + /DISCARD/ : {
> + *(.comment)
> + *(.comment.*)
> + *(.note.*)
> + }
> /* Dynamic linkage sections. Collected simply so we can check they're empty. */
> .got : {
> *(.got)
> @@ -64,3 +94,4 @@ ASSERT(SIZEOF(.igot.plt) == 0, ".igot.plt non-empty")
> ASSERT(SIZEOF(.iplt) == 0, ".iplt non-empty")
> ASSERT(SIZEOF(.plt) == 0, ".plt non-empty")
> ASSERT(SIZEOF(.rel) == 0, "leftover relocations")
> +ASSERT(SIZEOF(.data) == 0, "we don't want data")
3mdeb are getting around to rebasing/resubmitting the Trenchboot work
(Intel TXT and AMD SKINIT) backing QubeOS Anti-Evil-Maid.
While most of the cleanup is proving very helpful (i.e. reducing their
work), the lack of .data was seen as likely to be a blocker. Thinking
about this more, I'm now fairly certain we do not need to exclude data.
This object executes during boot in 32bit flat unpaged mode (i.e. there
are no actual restrictions during execution), and because it's all
wrapped in .init.text, its "just code" to the outside world. This means
it does not impact R^X that we're trying to arrange for the EFI section
headers.
Therefore the data arrangements should stay as they were before, and I
think the result will be fine. We obviously don't want gratuitous use
of .data, but we don't need to prohibit it either.
I've got various other suggestions for improvements of the result, but
they can all be deferred until later. This is complicated enough.
> diff --git a/xen/tools/combine_two_binaries b/xen/tools/combine_two_binaries
> new file mode 100755
> index 0000000000..ea2d6ddc4e
> --- /dev/null
> +++ b/xen/tools/combine_two_binaries
> @@ -0,0 +1,198 @@
> +#!/usr/bin/env python3
> +
> +from __future__ import print_function
> +import argparse
> +import re
> +import struct
> +import sys
> +
> +parser = argparse.ArgumentParser(description='Generate assembly file to merge into other code.')
> +parser.add_argument('--script', dest='script',
> + required=True,
> + help='Linker script for extracting symbols')
> +parser.add_argument('--bin1', dest='bin1',
> + required=True,
> + help='First binary')
> +parser.add_argument('--bin2', dest='bin2',
> + required=True,
> + help='Second binary')
> +parser.add_argument('--output', dest='output',
> + help='Output file')
> +parser.add_argument('--map', dest='mapfile',
> + help='Map file to read for symbols to export')
> +parser.add_argument('--exports', dest='exports',
> + help='Symbols to export')
> +parser.add_argument('--section-header', dest='section_header',
> + default='.text',
> + help='Section header declaration')
> +parser.add_argument('-v', '--verbose',
> + action='store_true')
> +args = parser.parse_args()
> +
> +gap = 0x010200
> +text_diff = 0x408020
> +
> +# Parse linker script for external symbols to use.
> +symbol_re = re.compile(r'\s+(\S+)\s*=\s*\.\s*\+\s*\((\d+)\s*\*\s*0\s*\)\s*;')
What is this looking for? I'd expect the DECLARE_IMPORT() lines, but
this is as clear as regexes...
> +symbols = {}
> +lines = {}
> +for line in open(args.script):
> + m = symbol_re.match(line)
> + if not m:
> + continue
> + (name, line_num) = (m.group(1), int(m.group(2)))
> + if line_num == 0:
> + raise Exception("Invalid line number found:\n\t" + line)
> + if line_num in symbols:
> + raise Exception("Symbol with this line already present:\n\t" + line)
> + if name in lines:
> + raise Exception("Symbol with this name already present:\n\t" + name)
> + symbols[line_num] = name
> + lines[name] = line_num
> +
> +exports = []
> +if args.exports is not None:
> + exports = dict([(name, None) for name in args.exports.split(',')])
> +
> +# Parse mapfile, look for ther symbols we want to export.
> +if args.mapfile is not None:
> + symbol_re = re.compile(r'\s{15,}0x([0-9a-f]+)\s+(\S+)\n')
> + for line in open(args.mapfile):
> + m = symbol_re.match(line)
> + if not m or m.group(2) not in exports:
> + continue
> + addr = int(m.group(1), 16)
> + exports[m.group(2)] = addr
> +for (name, addr) in exports.items():
> + if addr is None:
> + raise Exception("Required export symbols %s not found" % name)
> +
> +file1 = open(args.bin1, 'rb')
> +file2 = open(args.bin2, 'rb')
> +file1.seek(0, 2)
> +size1 = file1.tell()
> +file2.seek(0, 2)
> +size2 = file2.tell()
> +if size1 > size2:
> + file1, file2 = file2, file1
> + size1, size2 = size2, size1
> +if size2 != size1 + gap:
> + raise Exception('File sizes do not match')
> +
> +file1.seek(0, 0)
> +data1 = file1.read(size1)
> +file2.seek(gap, 0)
> +data2 = file2.read(size1)
> +
> +max_line = max(symbols.keys())
> +
> +def to_int32(n):
> + '''Convert a number to signed 32 bit integer truncating if needed'''
> + mask = (1 << 32) - 1
> + h = 1 << 31
> + return (n & mask) ^ h - h
> +
> +i = 0
> +references = {}
> +internals = 0
> +while i <= size1 - 4:
> + n1 = struct.unpack('<I', data1[i:i+4])[0]
> + n2 = struct.unpack('<I', data2[i:i+4])[0]
> + i += 1
> + # The two numbers are the same, no problems
> + if n1 == n2:
> + continue
> + # Try to understand why they are different
> + diff = to_int32(n1 - n2)
> + if diff == -gap: # this is an internal relocation
> + pos = i - 1
> + print(("Internal relocation found at position %#x "
> + "n1=%#x n2=%#x diff=%#x") % (pos, n1, n2, diff),
Here and elsewhere, you don't need brackets around around the string
itself. Python strings are like C strings and will auto-concatenate.
> + file=sys.stderr)
> + i += 3
> + internals += 1
> + if internals >= 10:
> + break
> + continue
> + # This is a relative relocation to a symbol, accepted, code/data is
> + # relocatable.
> + if diff < gap and diff >= gap - max_line:
> + n = gap - diff
> + symbol = symbols.get(n)
> + # check we have a symbol
> + if symbol is None:
> + raise Exception("Cannot find symbol for line %d" % n)
> + pos = i - 1
> + if args.verbose:
> + print('Position %#x %d %s' % (pos, n, symbol), file=sys.stderr)
> + i += 3
> + references[pos] = symbol
> + continue
> + # First byte is the same, move to next byte
> + if diff & 0xff == 0 and i <= size1 - 4:
> + continue
> + # Probably a type of relocation we don't want or support
> + pos = i - 1
> + suggestion = ''
> + symbol = symbols.get(-diff - text_diff)
> + if symbol is not None:
> + suggestion = " Maybe %s is not defined as hidden?" % symbol
> + raise Exception(("Unexpected difference found at %#x "
> + "n1=%#x n2=%#x diff=%#x gap=%#x.%s") % \
> + (pos, n1, n2, diff, gap, suggestion))
> +if internals != 0:
> + raise Exception("Previous relocations found")
> +
> +def line_bytes(buf, out):
> + '''Output an assembly line with all bytes in "buf"'''
> + if type(buf) == str:
> + print("\t.byte " + ','.join([str(ord(c)) for c in buf]), file=out)
> + else:
> + print("\t.byte " + ','.join([str(n) for n in buf]), file=out)
I'm guessing this is Py2/3 compatibility?
> +
> +def part(start, end, out):
> + '''Output bytes of "data" from "start" to "end"'''
> + while start < end:
> + e = min(start + 16, end)
> + line_bytes(data1[start:e], out)
> + start = e
> +
> +def reference(pos, out):
> + name = references[pos]
> + n = struct.unpack('<I', data1[pos:pos+4])[0]
> + sign = '+'
> + if n >= (1 << 31):
> + n -= (1 << 32)
> + n += pos
> + if n < 0:
> + n = -n
> + sign = '-'
> + print("\t.hidden %s\n\t.long %s %s %#x - ." % (name, name, sign, n),
Personally, I think this is easier to read as:
print("\t.hidden %s\n"
"\t.long %s %s %#x - ." % (name, name, sign, n),
file=out)
so it visually matches the output too. Same for .globl/hidden lower.
~Andrew
On Fri, Oct 11, 2024 at 2:17 PM Andrew Cooper <andrew.cooper3@citrix.com> wrote:
>
> On 11/10/2024 9:52 am, Frediano Ziglio wrote:
> > diff --git a/xen/arch/x86/boot/Makefile b/xen/arch/x86/boot/Makefile
> > index ff0f965876..4cf0d7e140 100644
> > --- a/xen/arch/x86/boot/Makefile
> > +++ b/xen/arch/x86/boot/Makefile
> > ...
> > +$(obj)/built_in_32.o: $(obj)/built_in_32.other.bin $(obj)/built_in_32.final.bin
> > + $(PYTHON) $(srctree)/tools/combine_two_binaries \
> > + --script $(obj)/build32.final.lds \
> > + --bin1 $(obj)/built_in_32.other.bin --bin2 $(obj)/built_in_32.final.bin \
> > + --map $(obj)/built_in_32.final.map \
> > + --exports cmdline_parse_early,reloc \
> > + --section-header '.section .init.text, "ax", @progbits' \
> > + --output $(obj)/built_in_32.s
>
> I can't see a case where we'd want this anywhere other than .init.text,
> so I'd drop the --section-header and just write it out unconditionally.
Could we just change the default in Python code and remove the option
calling the script?
> However, looking at the output:
>
> xen$ head arch/x86/boot/built_in_32.S
> .section .init.text, "ax", @progbits
> .byte 137,194,128,56,0,116,6,64,128,56,0,117,250,41,208,195
> .byte 133,201,116,42,86,83,141,52,8,64,15,182,72,255,66,15
> .byte 182,90,255,56,217,117,15,132,201,116,25,57,198,117,234,184
>
> This wants to start with a comment saying that it was automatically
> generated by combine_two_binaries.
>
Added a
print('''/*
* File autogenerated by combine_two_binaries.py DO NOT EDIT
*/''', file=out)
statement
And renamed the script file adding the ".py".
> Next, we need some kind of symbol at the start. Right now, the
> disassembly reads:
>
> arch/x86/boot/built_in_32.o: file format elf64-x86-64
> Disassembly of section .init.text:
> 0000000000000000 <cmdline_parse_early-0x391>:
> 0: 89 c2 mov %eax,%edx
> 2: 80 38 00 cmpb $0x0,(%eax)
>
>
> because most metadata is lost by this transformation, and it doesn't
> know that this is in fact strlen(). I'd suggest suggest obj32_start: or
> equivalent.
>
Would "obj_start" fine too? Maybe in the future it could be used for
64 bit too (to avoid mistakes like the relocation of error strings we
had).
> > diff --git a/xen/arch/x86/boot/build32.lds b/xen/arch/x86/boot/build32.lds.S
> > similarity index 63%
> > rename from xen/arch/x86/boot/build32.lds
> > rename to xen/arch/x86/boot/build32.lds.S
> > index 56edaa727b..72a4c5c614 100644
> > --- a/xen/arch/x86/boot/build32.lds
> > +++ b/xen/arch/x86/boot/build32.lds.S
> > @@ -15,22 +15,52 @@
> > * with this program. If not, see <http://www.gnu.org/licenses/>.
> > */
> >
> > -ENTRY(_start)
> > +#ifdef FINAL
> > +# define GAP 0
> > +# define MULT 0
> > +# define TEXT_START
> > +#else
> > +# define GAP 0x010200
> > +# define MULT 1
> > +# define TEXT_START 0x408020
> > +#endif
> > +# define DECLARE_IMPORT(name) name = . + (__LINE__ * MULT)
> > +
> > +ENTRY(dummy_start)
> >
> > SECTIONS
> > {
> > - /* Merge code and data into one section. */
> > - .text : {
> > + /* Merge code and read-only data into one section. */
> > + .text TEXT_START : {
> > + /* Silence linker warning, we are not going to use it */
> > + dummy_start = .;
> > +
> > + /* Declare below any symbol name needed.
> > + * Each symbol should be on its own line.
> > + * It looks like a tedious work but we make sure the things we use.
> > + * Potentially they should be all variables. */
> > + DECLARE_IMPORT(__base_relocs_start);
> > + DECLARE_IMPORT(__base_relocs_end);
> > + . = . + GAP;
> > *(.text)
> > *(.text.*)
> > - *(.data)
> > - *(.data.*)
> > *(.rodata)
> > *(.rodata.*)
> > + }
> > +
> > + /* Writeable data sections. Check empty.
> > + * We collapse all into code section and we don't want it to be writeable. */
> > + .data : {
> > + *(.data)
> > + *(.data.*)
> > *(.bss)
> > *(.bss.*)
> > }
> > -
> > + /DISCARD/ : {
> > + *(.comment)
> > + *(.comment.*)
> > + *(.note.*)
> > + }
> > /* Dynamic linkage sections. Collected simply so we can check they're empty. */
> > .got : {
> > *(.got)
> > @@ -64,3 +94,4 @@ ASSERT(SIZEOF(.igot.plt) == 0, ".igot.plt non-empty")
> > ASSERT(SIZEOF(.iplt) == 0, ".iplt non-empty")
> > ASSERT(SIZEOF(.plt) == 0, ".plt non-empty")
> > ASSERT(SIZEOF(.rel) == 0, "leftover relocations")
> > +ASSERT(SIZEOF(.data) == 0, "we don't want data")
>
> 3mdeb are getting around to rebasing/resubmitting the Trenchboot work
> (Intel TXT and AMD SKINIT) backing QubeOS Anti-Evil-Maid.
>
> While most of the cleanup is proving very helpful (i.e. reducing their
> work), the lack of .data was seen as likely to be a blocker. Thinking
> about this more, I'm now fairly certain we do not need to exclude data.
>
We could change if needed in the future.
Can I order:
- .text
- .rodata
- .data
- .bss
instead of old
- .text
- .data
- .rodata
- .bss
So all readonly code/data are more close?
> This object executes during boot in 32bit flat unpaged mode (i.e. there
> are no actual restrictions during execution), and because it's all
> wrapped in .init.text, its "just code" to the outside world. This means
> it does not impact R^X that we're trying to arrange for the EFI section
> headers.
>
> Therefore the data arrangements should stay as they were before, and I
> think the result will be fine. We obviously don't want gratuitous use
> of .data, but we don't need to prohibit it either.
>
> I've got various other suggestions for improvements of the result, but
> they can all be deferred until later. This is complicated enough.
>
> > diff --git a/xen/tools/combine_two_binaries b/xen/tools/combine_two_binaries
> > new file mode 100755
> > index 0000000000..ea2d6ddc4e
> > --- /dev/null
> > +++ b/xen/tools/combine_two_binaries
> > @@ -0,0 +1,198 @@
> > +#!/usr/bin/env python3
> > +
> > +from __future__ import print_function
> > +import argparse
> > +import re
> > +import struct
> > +import sys
> > +
> > +parser = argparse.ArgumentParser(description='Generate assembly file to merge into other code.')
> > +parser.add_argument('--script', dest='script',
> > + required=True,
> > + help='Linker script for extracting symbols')
> > +parser.add_argument('--bin1', dest='bin1',
> > + required=True,
> > + help='First binary')
> > +parser.add_argument('--bin2', dest='bin2',
> > + required=True,
> > + help='Second binary')
> > +parser.add_argument('--output', dest='output',
> > + help='Output file')
> > +parser.add_argument('--map', dest='mapfile',
> > + help='Map file to read for symbols to export')
> > +parser.add_argument('--exports', dest='exports',
> > + help='Symbols to export')
> > +parser.add_argument('--section-header', dest='section_header',
> > + default='.text',
> > + help='Section header declaration')
> > +parser.add_argument('-v', '--verbose',
> > + action='store_true')
> > +args = parser.parse_args()
> > +
> > +gap = 0x010200
> > +text_diff = 0x408020
> > +
> > +# Parse linker script for external symbols to use.
> > +symbol_re = re.compile(r'\s+(\S+)\s*=\s*\.\s*\+\s*\((\d+)\s*\*\s*0\s*\)\s*;')
>
> What is this looking for? I'd expect the DECLARE_IMPORT() lines, but
> this is as clear as regexes...
>
Added a
# Next regex matches expanded DECLARE_IMPORT lines in linker script.
comment
> > +symbols = {}
> > +lines = {}
> > +for line in open(args.script):
> > + m = symbol_re.match(line)
> > + if not m:
> > + continue
> > + (name, line_num) = (m.group(1), int(m.group(2)))
> > + if line_num == 0:
> > + raise Exception("Invalid line number found:\n\t" + line)
> > + if line_num in symbols:
> > + raise Exception("Symbol with this line already present:\n\t" + line)
> > + if name in lines:
> > + raise Exception("Symbol with this name already present:\n\t" + name)
> > + symbols[line_num] = name
> > + lines[name] = line_num
> > +
> > +exports = []
> > +if args.exports is not None:
> > + exports = dict([(name, None) for name in args.exports.split(',')])
> > +
> > +# Parse mapfile, look for ther symbols we want to export.
> > +if args.mapfile is not None:
> > + symbol_re = re.compile(r'\s{15,}0x([0-9a-f]+)\s+(\S+)\n')
> > + for line in open(args.mapfile):
> > + m = symbol_re.match(line)
> > + if not m or m.group(2) not in exports:
> > + continue
> > + addr = int(m.group(1), 16)
> > + exports[m.group(2)] = addr
> > +for (name, addr) in exports.items():
> > + if addr is None:
> > + raise Exception("Required export symbols %s not found" % name)
> > +
> > +file1 = open(args.bin1, 'rb')
> > +file2 = open(args.bin2, 'rb')
> > +file1.seek(0, 2)
> > +size1 = file1.tell()
> > +file2.seek(0, 2)
> > +size2 = file2.tell()
> > +if size1 > size2:
> > + file1, file2 = file2, file1
> > + size1, size2 = size2, size1
> > +if size2 != size1 + gap:
> > + raise Exception('File sizes do not match')
> > +
> > +file1.seek(0, 0)
> > +data1 = file1.read(size1)
> > +file2.seek(gap, 0)
> > +data2 = file2.read(size1)
> > +
> > +max_line = max(symbols.keys())
> > +
> > +def to_int32(n):
> > + '''Convert a number to signed 32 bit integer truncating if needed'''
> > + mask = (1 << 32) - 1
> > + h = 1 << 31
> > + return (n & mask) ^ h - h
> > +
> > +i = 0
> > +references = {}
> > +internals = 0
> > +while i <= size1 - 4:
> > + n1 = struct.unpack('<I', data1[i:i+4])[0]
> > + n2 = struct.unpack('<I', data2[i:i+4])[0]
> > + i += 1
> > + # The two numbers are the same, no problems
> > + if n1 == n2:
> > + continue
> > + # Try to understand why they are different
> > + diff = to_int32(n1 - n2)
> > + if diff == -gap: # this is an internal relocation
> > + pos = i - 1
> > + print(("Internal relocation found at position %#x "
> > + "n1=%#x n2=%#x diff=%#x") % (pos, n1, n2, diff),
>
> Here and elsewhere, you don't need brackets around around the string
> itself. Python strings are like C strings and will auto-concatenate.
>
Done (also another similar occurency below).
> > + file=sys.stderr)
> > + i += 3
> > + internals += 1
> > + if internals >= 10:
> > + break
> > + continue
> > + # This is a relative relocation to a symbol, accepted, code/data is
> > + # relocatable.
> > + if diff < gap and diff >= gap - max_line:
> > + n = gap - diff
> > + symbol = symbols.get(n)
> > + # check we have a symbol
> > + if symbol is None:
> > + raise Exception("Cannot find symbol for line %d" % n)
> > + pos = i - 1
> > + if args.verbose:
> > + print('Position %#x %d %s' % (pos, n, symbol), file=sys.stderr)
> > + i += 3
> > + references[pos] = symbol
> > + continue
> > + # First byte is the same, move to next byte
> > + if diff & 0xff == 0 and i <= size1 - 4:
> > + continue
> > + # Probably a type of relocation we don't want or support
> > + pos = i - 1
> > + suggestion = ''
> > + symbol = symbols.get(-diff - text_diff)
> > + if symbol is not None:
> > + suggestion = " Maybe %s is not defined as hidden?" % symbol
> > + raise Exception(("Unexpected difference found at %#x "
> > + "n1=%#x n2=%#x diff=%#x gap=%#x.%s") % \
> > + (pos, n1, n2, diff, gap, suggestion))
Here removed other parenthesis
> > +if internals != 0:
> > + raise Exception("Previous relocations found")
> > +
> > +def line_bytes(buf, out):
> > + '''Output an assembly line with all bytes in "buf"'''
> > + if type(buf) == str:
> > + print("\t.byte " + ','.join([str(ord(c)) for c in buf]), file=out)
> > + else:
> > + print("\t.byte " + ','.join([str(n) for n in buf]), file=out)
>
> I'm guessing this is Py2/3 compatibility?
>
Yes, as added
# Python 2 compatibility
will state
> > +
> > +def part(start, end, out):
> > + '''Output bytes of "data" from "start" to "end"'''
> > + while start < end:
> > + e = min(start + 16, end)
> > + line_bytes(data1[start:e], out)
> > + start = e
> > +
> > +def reference(pos, out):
> > + name = references[pos]
> > + n = struct.unpack('<I', data1[pos:pos+4])[0]
> > + sign = '+'
> > + if n >= (1 << 31):
> > + n -= (1 << 32)
> > + n += pos
> > + if n < 0:
> > + n = -n
> > + sign = '-'
> > + print("\t.hidden %s\n\t.long %s %s %#x - ." % (name, name, sign, n),
>
> Personally, I think this is easier to read as:
>
> print("\t.hidden %s\n"
> "\t.long %s %s %#x - ." % (name, name, sign, n),
> file=out)
>
> so it visually matches the output too. Same for .globl/hidden lower.
>
Done
> ~Andrew
Frediano
On 11/10/2024 9:52 am, Frediano Ziglio wrote: > <snip> > > Signed-off-by: Frediano Ziglio <frediano.ziglio@cloud.com> The makefile changes here are not the easiest to follow, because there are two related things being done. I experimented, and came up with the following: https://xenbits.xen.org/gitweb/?p=people/andrewcoop/xen.git;a=shortlog;h=refs/heads/fz-32b-v3.1 which splits the obj32 transformation out of the "totally change how we link things" which is the purpose of this patch. The end result is much clearer to follow IMO. I've got some other comments, which I've included in the branch, but are presented here for posterity. > diff --git a/xen/arch/x86/boot/Makefile b/xen/arch/x86/boot/Makefile > index ff0f965876..4cf0d7e140 100644 > --- a/xen/arch/x86/boot/Makefile > +++ b/xen/arch/x86/boot/Makefile > @@ -1,15 +1,18 @@ > +obj32 := cmdline.o > +obj32 += reloc.o > + > obj-bin-y += head.o > +obj-bin-y += built_in_32.o > > -head-bin-objs := cmdline.o reloc.o > +obj32 := $(patsubst %.o,%.32.o,$(obj32)) We're already used to writing out foo.init.o, so just have the .32.o in list. This makes patch 3 slightly more obvious, where you're not listing the same .o in two different obj lists. > > -nocov-y += $(head-bin-objs) > -noubsan-y += $(head-bin-objs) > -targets += $(head-bin-objs) > +nocov-y += $(obj32) > +noubsan-y += $(obj32) > +targets += $(obj32) > > -head-bin-objs := $(addprefix $(obj)/,$(head-bin-objs)) > +obj32 := $(addprefix $(obj)/,$(obj32)) > > $(obj)/head.o: AFLAGS-y += -Wa$(comma)-I$(obj) > -$(obj)/head.o: $(head-bin-objs:.o=.bin) The AFLAGS-y was only for the .incbin's, so can go away here too. > > CFLAGS_x86_32 := $(subst -m64,-m32 -march=i686,$(XEN_TREEWIDE_CFLAGS)) > $(call cc-options-add,CFLAGS_x86_32,CC,$(EMBEDDED_EXTRA_CFLAGS)) > @@ -17,17 +20,46 @@ CFLAGS_x86_32 += -Werror -fno-builtin -g0 -msoft-float -mregparm=3 > CFLAGS_x86_32 += -nostdinc -include $(filter %/include/xen/config.h,$(XEN_CFLAGS)) > CFLAGS_x86_32 += $(filter -I% -O%,$(XEN_CFLAGS)) -D__XEN__ > > +LD32 := $(LD) $(subst x86_64,i386,$(LDFLAGS_DIRECT)) Because this is :=, it needs to be after ... > + > # override for 32bit binaries > -$(head-bin-objs): CFLAGS_stack_boundary := > -$(head-bin-objs): XEN_CFLAGS := $(CFLAGS_x86_32) -fpic > +$(obj32): CFLAGS_stack_boundary := > +$(obj32): XEN_CFLAGS := $(CFLAGS_x86_32) -fpic > > LDFLAGS_DIRECT-$(call ld-option,--warn-rwx-segments) := --no-warn-rwx-segments > LDFLAGS_DIRECT += $(LDFLAGS_DIRECT-y) ... this. > > -%.bin: %.lnk > - $(OBJCOPY) -j .text -O binary $< $@ > +$(obj)/%.32.o: $(src)/%.c FORCE > + $(call if_changed_rule,cc_o_c) > + > +$(obj)/build32.final.lds: AFLAGS-y += -DFINAL > +$(obj)/build32.other.lds $(obj)/build32.final.lds: $(src)/build32.lds.S > + $(call if_changed_dep,cpp_lds_S) FORCE needs to be a prerequisite to use $(call if_changed_dep) (Sorry - I missed this one in the branch I pushed.) > + > +orphan-handling-$(call ld-option,--orphan-handling=error) := --orphan-handling=error > + > +# link all object files together All 32bit objects. It isn't even all objects today (head.o isn't included). > +$(obj)/built_in_32.tmp.o: $(obj32) > + $(LD32) -r -o $@ $(obj32) $^ > + > +$(obj)/built_in_32.%.bin: $(obj)/build32.%.lds $(obj)/built_in_32.tmp.o > +## link bundle with a given layout > + $(LD32) $(orphan-handling-y) -N -T $< -Map $(obj)/built_in_32.$(*F).map -o $(obj)/built_in_32.$(*F).o $(obj)/built_in_32.tmp.o > +## extract binaries from object > + $(OBJCOPY) -j .text -O binary $(obj)/built_in_32.$(*F).o $@ > + rm -f $(obj)/built_in_32.$(*F).o > > -%.lnk: %.o $(src)/build32.lds > - $(LD) $(subst x86_64,i386,$(LDFLAGS_DIRECT)) -N -T $(filter %.lds,$^) -o $@ $< > +# generate final object file combining and checking above binaries > +$(obj)/built_in_32.o: $(obj)/built_in_32.other.bin $(obj)/built_in_32.final.bin > + $(PYTHON) $(srctree)/tools/combine_two_binaries \ > + --script $(obj)/build32.final.lds \ > + --bin1 $(obj)/built_in_32.other.bin --bin2 $(obj)/built_in_32.final.bin \ > + --map $(obj)/built_in_32.final.map \ > + --exports cmdline_parse_early,reloc \ > + --section-header '.section .init.text, "ax", @progbits' \ > + --output $(obj)/built_in_32.s > + $(CC) -c $(obj)/built_in_32.s -o $@.tmp > + rm -f $(obj)/built_in_32.s $@ > + mv $@.tmp $@ $(obj)/built_in_32.S: and --output $@ The build system already knows how to turn a .S into a .o. This form removed an opencoded $(CC), and leaves built_in_32.S around as an intermediate to be inspected. ~Andrew
© 2016 - 2024 Red Hat, Inc.