| Message ID | 20221107223921.3451913-6-song@kernel.org (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | execmem_alloc for BPF programs | expand |
On Mon, 2022-11-07 at 14:39 -0800, Song Liu wrote: > Allocate 2MB pages up to round_up(_etext, 2MB), and register memory > [round_up(_etext, 4kb), round_up(_etext, 2MB)] with > register_text_tail_vm > so that we can use this part of memory for dynamic kernel text (BPF > programs, etc.). > > Here is an example: > > [root@eth50-1 ~]# grep _etext /proc/kallsyms > ffffffff82202a08 T _etext > > [root@eth50-1 ~]# grep bpf_prog_ /proc/kallsyms | tail -n 3 > ffffffff8220f920 t > bpf_prog_cc61a5364ac11d93_handle__sched_wakeup [bpf] > ffffffff8220fa28 t > bpf_prog_cc61a5364ac11d93_handle__sched_wakeup_new [bpf] > ffffffff8220fad4 t > bpf_prog_3bf73fa16f5e3d92_handle__sched_switch [bpf] > > [root@eth50-1 ~]# grep 0xffffffff82200000 > /sys/kernel/debug/page_tables/kernel > 0xffffffff82200000-0xffffffff82400000 2M ro PSE x > pmd > > ffffffff82200000-ffffffff82400000 is a 2MB page, serving kernel text, > and > bpf programs. > > Signed-off-by: Song Liu <song@kernel.org> Please update Documentation/x86/x86_64/mm.txt and teach places that check if an address is text about it.
On Tue, Nov 8, 2022 at 11:04 AM Edgecombe, Rick P <rick.p.edgecombe@intel.com> wrote: > > On Mon, 2022-11-07 at 14:39 -0800, Song Liu wrote: > > Allocate 2MB pages up to round_up(_etext, 2MB), and register memory > > [round_up(_etext, 4kb), round_up(_etext, 2MB)] with > > register_text_tail_vm > > so that we can use this part of memory for dynamic kernel text (BPF > > programs, etc.). > > > > Here is an example: > > > > [root@eth50-1 ~]# grep _etext /proc/kallsyms > > ffffffff82202a08 T _etext > > > > [root@eth50-1 ~]# grep bpf_prog_ /proc/kallsyms | tail -n 3 > > ffffffff8220f920 t > > bpf_prog_cc61a5364ac11d93_handle__sched_wakeup [bpf] > > ffffffff8220fa28 t > > bpf_prog_cc61a5364ac11d93_handle__sched_wakeup_new [bpf] > > ffffffff8220fad4 t > > bpf_prog_3bf73fa16f5e3d92_handle__sched_switch [bpf] > > > > [root@eth50-1 ~]# grep 0xffffffff82200000 > > /sys/kernel/debug/page_tables/kernel > > 0xffffffff82200000-0xffffffff82400000 2M ro PSE x > > pmd > > > > ffffffff82200000-ffffffff82400000 is a 2MB page, serving kernel text, > > and > > bpf programs. > > > > Signed-off-by: Song Liu <song@kernel.org> > > Please update Documentation/x86/x86_64/mm.txt and teach places that > check if an address is text about it. For mm.rst, I got something like: =========================== 8< =========================== diff --git i/Documentation/x86/x86_64/mm.rst w/Documentation/x86/x86_64/mm.rst index 9798676bb0bf..ac041b7d3965 100644 --- i/Documentation/x86/x86_64/mm.rst +++ w/Documentation/x86/x86_64/mm.rst @@ -62,7 +62,7 @@ Complete virtual memory map with 4-level page tables ffffff8000000000 | -512 GB | ffffffeeffffffff | 444 GB | ... unused hole ffffffef00000000 | -68 GB | fffffffeffffffff | 64 GB | EFI region mapping space ffffffff00000000 | -4 GB | ffffffff7fffffff | 2 GB | ... unused hole - ffffffff80000000 | -2 GB | ffffffff9fffffff | 512 MB | kernel text mapping, mapped to physical address 0 + ffffffff80000000 | -2 GB | ffffffff9fffffff | 512 MB | kernel and module text mapping, mapped to physical address 0 ffffffff80000000 |-2048 MB | | | ffffffffa0000000 |-1536 MB | fffffffffeffffff | 1520 MB | module mapping space ffffffffff000000 | -16 MB | | | @@ -121,7 +121,7 @@ Complete virtual memory map with 5-level page tables ffffff8000000000 | -512 GB | ffffffeeffffffff | 444 GB | ... unused hole ffffffef00000000 | -68 GB | fffffffeffffffff | 64 GB | EFI region mapping space ffffffff00000000 | -4 GB | ffffffff7fffffff | 2 GB | ... unused hole - ffffffff80000000 | -2 GB | ffffffff9fffffff | 512 MB | kernel text mapping, mapped to physical address 0 + ffffffff80000000 | -2 GB | ffffffff9fffffff | 512 MB | kernel and module text mapping, mapped to physical address 0 ffffffff80000000 |-2048 MB | | | ffffffffa0000000 |-1536 MB | fffffffffeffffff | 1520 MB | module mapping space ffffffffff000000 | -16 MB | | | =========================== 8< =========================== Is this good enough? I added extra check in is_vmalloc_or_module_addr() (4/5). Where do we need similar logic? Thanks, Song
On Tue, 2022-11-08 at 14:15 -0800, Song Liu wrote: > diff --git i/Documentation/x86/x86_64/mm.rst > w/Documentation/x86/x86_64/mm.rst > index 9798676bb0bf..ac041b7d3965 100644 > --- i/Documentation/x86/x86_64/mm.rst > +++ w/Documentation/x86/x86_64/mm.rst > @@ -62,7 +62,7 @@ Complete virtual memory map with 4-level page > tables > ffffff8000000000 | -512 GB | ffffffeeffffffff | 444 GB | ... > unused hole > ffffffef00000000 | -68 GB | fffffffeffffffff | 64 GB | EFI > region mapping space > ffffffff00000000 | -4 GB | ffffffff7fffffff | 2 GB | ... > unused hole > - ffffffff80000000 | -2 GB | ffffffff9fffffff | 512 MB | > kernel text mapping, mapped to physical address 0 > + ffffffff80000000 | -2 GB | ffffffff9fffffff | 512 MB | > kernel and module text mapping, mapped to physical address 0 It's not really "module text mapping" yet right? Because it doesn't get used by modules. I might just call it execmem or whatever you call the component. Otherwise it is outdated when the next users starts using the API. Otherwise looks ok, thanks.
diff --git a/arch/x86/include/asm/pgtable_64_types.h b/arch/x86/include/asm/pgtable_64_types.h index 04f36063ad54..c0f9cceb109a 100644 --- a/arch/x86/include/asm/pgtable_64_types.h +++ b/arch/x86/include/asm/pgtable_64_types.h @@ -101,6 +101,7 @@ extern unsigned int ptrs_per_p4d; #define PUD_MASK (~(PUD_SIZE - 1)) #define PGDIR_SIZE (_AC(1, UL) << PGDIR_SHIFT) #define PGDIR_MASK (~(PGDIR_SIZE - 1)) +#define PMD_ALIGN(x) (((unsigned long)(x) + (PMD_SIZE - 1)) & PMD_MASK) /* * See Documentation/x86/x86_64/mm.rst for a description of the memory map. diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c index 3f040c6e5d13..5b42fc0c6099 100644 --- a/arch/x86/mm/init_64.c +++ b/arch/x86/mm/init_64.c @@ -1373,7 +1373,7 @@ void mark_rodata_ro(void) unsigned long start = PFN_ALIGN(_text); unsigned long rodata_start = PFN_ALIGN(__start_rodata); unsigned long end = (unsigned long)__end_rodata_hpage_align; - unsigned long text_end = PFN_ALIGN(_etext); + unsigned long text_end = PMD_ALIGN(_etext); unsigned long rodata_end = PFN_ALIGN(__end_rodata); unsigned long all_end; @@ -1414,6 +1414,8 @@ void mark_rodata_ro(void) (void *)rodata_end, (void *)_sdata); debug_checkwx(); + register_text_tail_vm(PFN_ALIGN((unsigned long)_etext), + PMD_ALIGN((unsigned long)_etext)); } int kern_addr_valid(unsigned long addr)
Allocate 2MB pages up to round_up(_etext, 2MB), and register memory [round_up(_etext, 4kb), round_up(_etext, 2MB)] with register_text_tail_vm so that we can use this part of memory for dynamic kernel text (BPF programs, etc.). Here is an example: [root@eth50-1 ~]# grep _etext /proc/kallsyms ffffffff82202a08 T _etext [root@eth50-1 ~]# grep bpf_prog_ /proc/kallsyms | tail -n 3 ffffffff8220f920 t bpf_prog_cc61a5364ac11d93_handle__sched_wakeup [bpf] ffffffff8220fa28 t bpf_prog_cc61a5364ac11d93_handle__sched_wakeup_new [bpf] ffffffff8220fad4 t bpf_prog_3bf73fa16f5e3d92_handle__sched_switch [bpf] [root@eth50-1 ~]# grep 0xffffffff82200000 /sys/kernel/debug/page_tables/kernel 0xffffffff82200000-0xffffffff82400000 2M ro PSE x pmd ffffffff82200000-ffffffff82400000 is a 2MB page, serving kernel text, and bpf programs. Signed-off-by: Song Liu <song@kernel.org> --- arch/x86/include/asm/pgtable_64_types.h | 1 + arch/x86/mm/init_64.c | 4 +++- 2 files changed, 4 insertions(+), 1 deletion(-)