Re: [RFC v2 4/4] vmalloc_exec: share a huge page with kernel text

[Song Liu]

> On Oct 10, 2022, at 11:32 AM, Edgecombe, Rick P <rick.p.edgecombe@xxxxxxxxx> wrote:
> 
> On Fri, 2022-10-07 at 16:43 -0700, Song Liu wrote:
>> On x86 kernel, we allocate 2MB pages for kernel text up to
>> round_down(_etext, 2MB). Therefore, some of the kernel text is still
>> on 4kB pages. With vmalloc_exec, we can allocate 2MB pages up to
>> round_up(_etext, 2MB), and use the rest of the page for modules and
>> BPF programs.
>> 
>> 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
>> 
>> [root@eth50-1 ~]# grep xfs_flush_inodes /proc/kallsyms
>> ffffffff822ba910 t xfs_flush_inodes_worker      [xfs]
>> ffffffff822bc580 t xfs_flush_inodes     [xfs]
>> 
>> ffffffff82200000-ffffffff82400000 is a 2MB page, serving kernel text,
>> xfs
>> module, and bpf programs.
> 
> Can this memory range be freed as part of a vfree_exec() call then?
> Does vmalloc actually try to unmap it? If so, it could get complicated
> with PTI.
> 
> It probably should be a special case that never gets fully freed.

Right, this is never freed. 

> 
>> 
>> Signed-off-by: Song Liu <song@xxxxxxxxxx>
>> ---
>> arch/x86/mm/init_64.c |  3 ++-
>> mm/vmalloc.c          | 24 ++++++++++++++++++++++++
>> 2 files changed, 26 insertions(+), 1 deletion(-)
>> 
>> diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
>> index 0fe690ebc269..d94f196c541a 100644
>> --- a/arch/x86/mm/init_64.c
>> +++ b/arch/x86/mm/init_64.c
>> @@ -1367,12 +1367,13 @@ int __init
>> deferred_page_init_max_threads(const struct cpumask *node_cpumask)
>> 
>> int kernel_set_to_readonly;
>> 
>> +#define PMD_ALIGN(x)	(((unsigned long)(x) + (PMD_SIZE - 1)) &
>> PMD_MASK)
>> 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);
> 
> This should probably have more logic and adjustments. If etext is PMD
> aligned, some of the stuff outside the diff won't do anything.

Hmm.. I don't quite follow this comment. If the etext is PMD aligned, 
we can still use vmalloc_exec to allocate memory. So it shouldn't 
matter, no?

> 
> Also, if a kernel doesn't have modules or BPF JIT it would be a waste
> of memory.

I guess we can add a command line argument for these corner cases? 

Thanks,
Song

> 
>> 	unsigned long rodata_end = PFN_ALIGN(__end_rodata);
>> 	unsigned long all_end;
>> 
>> diff --git a/mm/vmalloc.c b/mm/vmalloc.c
>> index 9212ff96b871..41509bbec583 100644
>> --- a/mm/vmalloc.c
>> +++ b/mm/vmalloc.c
>> @@ -75,6 +75,9 @@ static const bool vmap_allow_huge = false;
>> #define PMD_ALIGN(addr) ALIGN(addr, PMD_SIZE)
>> #define PMD_ALIGN_DOWN(addr) ALIGN_DOWN(addr, PMD_SIZE)
>> 
>> +static struct vm_struct text_tail_vm;
>> +static struct vmap_area text_tail_va;
>> +
>> bool is_vmalloc_addr(const void *x)
>> {
>> 	unsigned long addr = (unsigned long)kasan_reset_tag(x);
>> @@ -637,6 +640,8 @@ int is_vmalloc_or_module_addr(const void *x)
>> 	unsigned long addr = (unsigned long)kasan_reset_tag(x);
>> 	if (addr >= MODULES_VADDR && addr < MODULES_END)
>> 		return 1;
>> +	if (addr >= text_tail_va.va_start && addr <
>> text_tail_va.va_end)
>> +		return 1;
>> #endif
>> 	return is_vmalloc_addr(x);
>> }
>> @@ -2422,6 +2427,24 @@ static void vmap_init_free_space(void)
>> 	}
>> }
>> 
>> +static void register_text_tail_vm(void)
>> +{
>> +	unsigned long start = PFN_ALIGN((unsigned long)_etext);
>> +	unsigned long end = PMD_ALIGN((unsigned long)_etext);
>> +	struct vmap_area *va;
>> +
>> +	va = kmem_cache_zalloc(vmap_area_cachep, GFP_NOWAIT);
>> +	if (WARN_ON_ONCE(!va))
>> +		return;
>> +	text_tail_vm.addr = (void *)start;
>> +	text_tail_vm.size = end - start;
>> +	text_tail_va.va_start = start;
>> +	text_tail_va.va_end = end;
>> +	text_tail_va.vm = &text_tail_vm;
>> +	memcpy(va, &text_tail_va, sizeof(*va));
>> +	insert_vmap_area_augment(va, NULL, &free_text_area_root,
>> &free_text_area_list);
>> +}
>> +
>> void __init vmalloc_init(void)
>> {
>> 	struct vmap_area *va;
>> @@ -2432,6 +2455,7 @@ void __init vmalloc_init(void)
>> 	 * Create the cache for vmap_area objects.
>> 	 */
>> 	vmap_area_cachep = KMEM_CACHE(vmap_area, SLAB_PANIC);
>> +	register_text_tail_vm();
>> 
>> 	for_each_possible_cpu(i) {
>> 		struct vmap_block_queue *vbq;