diff options
| author | Pingfan Liu <piliu@redhat.com> | 2021-12-10 10:57:33 +0800 |
|---|---|---|
| committer | Simon Horman <horms@verge.net.au> | 2021-12-15 13:38:13 +0100 |
| commit | 1614959f2f0d79b5f7fdc57d6aa42da1a0298153 (patch) | |
| tree | 3fd01f76673bb32db8c5bb6c781fef3ada9ce1db /util_lib/elf_info.c | |
| parent | 10b7e1fb5d0cafb2051df56443ffb4bb79d34c7e (diff) | |
arm64/crashdump: deduce paddr of _text based on kernel code size
kexec-tools commit 61b8c79b0fb7 ("arm64/crashdump-arm64: deduce the
paddr of _text") tries to deduce the paddr of _text, but turns out
partially.
That commit is based on "The Image must be placed text_offset bytes from
a 2MB aligned base address anywhere in usable system RAM and called
there" in linux/Documentation/arm64/booting.rst, plus text_offset field
is zero.
But in practice, some boot loaders does not obey the convention, and
still boots up the kernel successfully.
Revisiting kernel commit e2a073dde921 ("arm64: omit [_text, _stext) from
permanent kernel mapping"), the kernel code size changes from (unsigned
long)__init_begin - (unsigned long)_text to (unsigned long)__init_begin
- (unsigned long)_stext
And it should be a better factor to decide which label starts the
"Kernel code" in /proc/iomem.
Signed-off-by: Pingfan Liu <piliu@redhat.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Diffstat (limited to 'util_lib/elf_info.c')
0 files changed, 0 insertions, 0 deletions