Message ID | 20241016111458.846228-1-kirill.shutemov@linux.intel.com (mailing list archive) |
---|---|
Headers | show |
Series | x86: Reduce code duplication on page table initialization | expand |
On Wed, Oct 16, 2024 at 02:14:54PM +0300, Kirill A. Shutemov wrote: > Use kernel_ident_mapping_init() to initialize kernel page tables where > possible, replacing manual initialization, reducing code duplication. > > v4: > - Reviewed-bys from Kai; > - Fix comment in acpi_mp_setup_reset() (Rafael); > v3: > - Reviewed-bys from Tom; > - Improve commit messages; > v2: > - A separate patch to change what PA is mapped at relocate_kernel() VA. > - Improve commit messages; > - Add Reveiwed-by from Kai; > > Kirill A. Shutemov (4): > x86/mm/ident_map: Fix virtual address wrap to zero > x86/acpi: Replace manual page table initialization with > kernel_ident_mapping_init() > x86/64/kexec: Map original relocate_kernel() in > init_transition_pgtable() > x86/64/kexec: Rewrite init_transition_pgtable() with > kernel_ident_mapping_init() > > arch/x86/include/asm/kexec.h | 5 +- > arch/x86/kernel/acpi/madt_wakeup.c | 73 +++++------------------- > arch/x86/kernel/machine_kexec_64.c | 89 +++++++++++------------------- > arch/x86/mm/ident_map.c | 14 +---- > 4 files changed, 50 insertions(+), 131 deletions(-) Any feedback on this series?
On 10/29/24 08:09, Kirill A. Shutemov wrote:
> Any feedback on this series?
It's tempting because it removes so much code, but it's also worrying
because these code paths are notoriously hard to test.
I'd be much happier if you'd send this after 6.13-rc1 drops and we give
it as much testing time as possible.
* Kirill A. Shutemov <kirill.shutemov@linux.intel.com> wrote: > Use kernel_ident_mapping_init() to initialize kernel page tables where > possible, replacing manual initialization, reducing code duplication. > > v4: > - Reviewed-bys from Kai; > - Fix comment in acpi_mp_setup_reset() (Rafael); > v3: > - Reviewed-bys from Tom; > - Improve commit messages; > v2: > - A separate patch to change what PA is mapped at relocate_kernel() VA. > - Improve commit messages; > - Add Reveiwed-by from Kai; > > Kirill A. Shutemov (4): > x86/mm/ident_map: Fix virtual address wrap to zero > x86/acpi: Replace manual page table initialization with kernel_ident_mapping_init() > x86/64/kexec: Map original relocate_kernel() in init_transition_pgtable() > x86/64/kexec: Rewrite init_transition_pgtable() with kernel_ident_mapping_init() > > arch/x86/include/asm/kexec.h | 5 +- > arch/x86/kernel/acpi/madt_wakeup.c | 73 +++++------------------- > arch/x86/kernel/machine_kexec_64.c | 89 +++++++++++------------------- > arch/x86/mm/ident_map.c | 14 +---- > 4 files changed, 50 insertions(+), 131 deletions(-) So looks like this series feel between the cracks during the holiday season. To help move them along, I've fixed up the first patch with the review feedback clarification requests, and applied patch #1 and #2 to tip:x86/mm: 4f10ec03fe1e ("x86/mm/ident_map: Fix theoretical virtual address overflow to zero") 376daf20eda4 ("x86/acpi: Replace manual page table initialization with kernel_ident_mapping_init()") Patches #3 and #4 don't apply anymore, due to interference by other work with commits like: 4b5bc2ec9a23 ("x86/kexec: Allocate PGD for x86_64 transition page tables separately") If the remaining patches are still relevant, mind porting them to latest -tip? Thanks, Ingo