Re: [PATCH v4][makedumpfile 7/7] Filter amdgpu mm pages

[Stephen Brennan <stephen.s.brennan@oracle.com>]

Tao Liu <ltao@redhat.com> writes:

> This patch will introduce maple_tree & amdgpu mm page filtering extension,
> those mm pages allocated to amdgpu will be discarded from vmcore, in order
> to shrink vmcore size since mm pages allocated to amdgpu are useless to kernel
> crash and may contain sensitive data.
>
> Signed-off-by: Tao Liu <ltao@redhat.com>
> ---
>  extensions/Makefile        |   4 +-
>  extensions/amdgpu_filter.c | 190 +++++++++++++++++++++++
>  extensions/maple_tree.c    | 307 +++++++++++++++++++++++++++++++++++++
>  extensions/maple_tree.h    |   6 +
>  4 files changed, 506 insertions(+), 1 deletion(-)
>  create mode 100644 extensions/amdgpu_filter.c
>  create mode 100644 extensions/maple_tree.c
>  create mode 100644 extensions/maple_tree.h
>
> diff --git a/extensions/Makefile b/extensions/Makefile
> index b8bbfbc..55b789b 100644
> --- a/extensions/Makefile
> +++ b/extensions/Makefile
> @@ -1,8 +1,10 @@
>  CC ?= gcc
> -CONTRIB_SO :=
> +CONTRIB_SO := amdgpu_filter.so
>  
>  all: $(CONTRIB_SO)
>  
> +amdgpu_filter.so: maple_tree.c
> +
>  $(CONTRIB_SO): %.so: %.c
>  	$(CC) -O2 -g -fPIC -shared -Wl,-T,../makedumpfile.ld -o $@ $^
>  
> diff --git a/extensions/amdgpu_filter.c b/extensions/amdgpu_filter.c
> new file mode 100644
> index 0000000..3a1e9f2
> --- /dev/null
> +++ b/extensions/amdgpu_filter.c
> @@ -0,0 +1,190 @@
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include "maple_tree.h"
> +#include "../makedumpfile.h"
> +#include "../btf_info.h"
> +#include "../kallsyms.h"
> +#include "../extension.h"
> +
> +/*
> + * These syms/types are must-have for the extension.
> +*/
> +INIT_KERN_STRUCT_MEMBER(task_struct, tasks);
> +INIT_KERN_STRUCT_MEMBER(task_struct, mm);
> +INIT_KERN_STRUCT_MEMBER(mm_struct, mm_mt);
> +INIT_KERN_STRUCT_MEMBER(vm_area_struct, vm_ops);
> +INIT_KERN_STRUCT_MEMBER(vm_area_struct, vm_private_data);
> +INIT_MOD_STRUCT_MEMBER(amdgpu, ttm_buffer_object, ttm);
> +INIT_MOD_STRUCT_MEMBER(amdgpu, ttm_tt, pages);
> +INIT_MOD_STRUCT_MEMBER(amdgpu, ttm_tt, num_pages);
> +INIT_KERN_STRUCT(page);
> +
> +INIT_KERN_SYM(init_task);
> +INIT_KERN_SYM(vmemmap_base);
> +INIT_MOD_SYM(amdgpu, amdgpu_gem_vm_ops);
> +
> +struct ft_page_info {
> +	unsigned long pfn;
> +	unsigned long num;
> +	struct ft_page_info *next;
> +};
> +
> +static struct ft_page_info *ft_head_discard = NULL;
> +
> +static void update_filter_pages_info(unsigned long pfn, unsigned long num)
> +{
> +	struct ft_page_info *p, **ft_head;
> +	struct ft_page_info *new_p = malloc(sizeof(struct ft_page_info));
> +
> +	ft_head = &ft_head_discard;
> +
> +	if (!new_p) {
> +		fprintf(stderr, "%s: Can't allocate memory for ft_page_info\n",
> +			__func__);
> +		return;
> +	}
> +	new_p->pfn = pfn;
> +	new_p->num = num;
> +	new_p->next = NULL;
> +
> +	if (!(*ft_head) || (*ft_head)->pfn > new_p->pfn) {
> +		new_p->next = (*ft_head);
> +		(*ft_head) = new_p;
> +		return;
> +	}
> +
> +	p = (*ft_head);
> +	while (p->next != NULL && p->next->pfn < new_p->pfn) {
> +		p = p->next;
> +	}
> +
> +	new_p->next = p->next;
> +	p->next = new_p;
> +}
> +
> +static int filter_page(unsigned long pfn, struct ft_page_info **p)
> +{
> +	struct ft_page_info *ft_head = ft_head_discard;
> +
> +	if (ft_head == NULL)
> +		return PG_UNDECID;
> +
> +	if (*p == NULL)
> +		*p = ft_head;
> +
> +	/* The gap before 1st block */
> +	if (pfn >= 0 && pfn < ft_head->pfn)
> +		return PG_UNDECID;
> +
> +	/* Handle 1~(n-1) blocks and following gaps */
> +	while ((*p)->next) {
> +		if (pfn >= (*p)->pfn && pfn < (*p)->pfn + (*p)->num)
> +			return PG_EXCLUDE; // hit the block
> +		if (pfn >= (*p)->pfn + (*p)->num && pfn < (*p)->next->pfn)
> +			return PG_UNDECID; // the gap after the block
> +		*p = (*p)->next;
> +	}
> +
> +	/* The last block and gap */
> +	if (pfn >= (*p)->pfn + (*p)->num)
> +		return PG_UNDECID;
> +	else
> +		return PG_EXCLUDE;
> +}
> +
> +static void do_cleanup(struct ft_page_info **ft_head)
> +{
> +	struct ft_page_info *p, *p_tmp;
> +
> +	for (p = *ft_head; p;) {
> +		p_tmp = p;
> +		p = p->next;
> +		free(p_tmp);
> +	}
> +	*ft_head = NULL;
> +}
> +
> +#define KERN_MEMBER_OFF(S, M) \
> +	GET_KERN_STRUCT_MEMBER_MOFF(S, M) / 8
> +#define MOD_MEMBER_OFF(MOD, S, M) \
> +	GET_MOD_STRUCT_MEMBER_MOFF(MOD, S, M) / 8
> +
> +static void gather_amdgpu_mm_range_info(void)
> +{
> +	uint64_t init_task, list, list_offset, amdgpu_gem_vm_ops;
> +	uint64_t mm, vm_ops, tbo, ttm, num_pages, pages, pfn, vmemmap_base;
> +	int array_len;
> +	unsigned long *array_out;
> +	init_task = GET_KERN_SYM(init_task);
> +	amdgpu_gem_vm_ops = GET_MOD_SYM(amdgpu, amdgpu_gem_vm_ops);
> +
> +	list = init_task + KERN_MEMBER_OFF(task_struct, tasks);
> +
> +	do {
> +		readmem(VADDR, list - KERN_MEMBER_OFF(task_struct, tasks) + 
> +				KERN_MEMBER_OFF(task_struct, mm),
> +			&mm, sizeof(uint64_t));
> +		if (!mm) {
> +			list = next_list(list);
> +			continue;
> +		}
> +
> +		array_out = mt_dump(mm + KERN_MEMBER_OFF(mm_struct, mm_mt), &array_len);
> +		if (!array_out)
> +			return;
> +
> +		for (int i = 0; i < array_len; i++) {
> +			num_pages = 0;
> +			readmem(VADDR, array_out[i] + KERN_MEMBER_OFF(vm_area_struct, vm_ops),
> +				&vm_ops, GET_KERN_STRUCT_MEMBER_MSIZE(vm_area_struct, vm_ops));
> +			if (vm_ops == amdgpu_gem_vm_ops) {
> +				readmem(VADDR, array_out[i] +
> +					KERN_MEMBER_OFF(vm_area_struct, vm_private_data),
> +					&tbo, GET_KERN_STRUCT_MEMBER_MSIZE(vm_area_struct, vm_private_data));
> +				readmem(VADDR, tbo + MOD_MEMBER_OFF(amdgpu, ttm_buffer_object, ttm),
> +					&ttm, GET_MOD_STRUCT_MEMBER_MSIZE(amdgpu, ttm_buffer_object, ttm));
> +				if (ttm) {
> +					readmem(VADDR, ttm + MOD_MEMBER_OFF(amdgpu, ttm_tt, num_pages),
> +						&num_pages, GET_MOD_STRUCT_MEMBER_MSIZE(amdgpu, ttm_tt, num_pages));
> +					readmem(VADDR, ttm + MOD_MEMBER_OFF(amdgpu, ttm_tt, pages),
> +						&pages, GET_MOD_STRUCT_MEMBER_MSIZE(amdgpu, ttm_tt, pages));

Are these pages guaranteed to be contiguous?
If not, you're applying the page filter rule based on just the first
element of the pages array.
If they are guaranteed contiguous, maybe a comment would be helpful?

Thanks,
Stephen

> +					readmem(VADDR, pages, &pages, sizeof(unsigned long));
> +					readmem(VADDR, GET_KERN_SYM(vmemmap_base),
> +						&vmemmap_base, sizeof(unsigned long));
> +					pfn = (pages - vmemmap_base) / GET_KERN_STRUCT_SSIZE(page);
> +					update_filter_pages_info(pfn, num_pages);
> +				}
> +			}
> +		}
> +
> +		free(array_out);
> +		list = next_list(list);
> +	} while (list != init_task + KERN_MEMBER_OFF(task_struct, tasks));
> +
> +	return;
> +}
> +
> +/* Extension callback when makedumpfile do page filtering */
> +int extension_callback(unsigned long pfn, const void *pcache)
> +{
> +	struct ft_page_info *cur = NULL;
> +
> +	return filter_page(pfn, &cur);
> +}
> +
> +/* Entry of extension */
> +void extension_init(void)
> +{
> +	if (!maple_init()) {
> +		goto out;
> +	}
> +	gather_amdgpu_mm_range_info();
> +out:
> +	return;
> +}
> +
> +__attribute__((destructor))
> +void extension_cleanup(void)
> +{
> +	do_cleanup(&ft_head_discard);
> +}
> diff --git a/extensions/maple_tree.c b/extensions/maple_tree.c
> new file mode 100644
> index 0000000..e367940
> --- /dev/null
> +++ b/extensions/maple_tree.c
> @@ -0,0 +1,307 @@
> +#include <stdio.h>
> +#include <stdbool.h>
> +#include "../btf_info.h"
> +#include "../kallsyms.h"
> +#include "../makedumpfile.h"
> +
> +static unsigned char mt_slots[4] = {0};
> +static unsigned char mt_pivots[4] = {0};
> +static unsigned long mt_max[4] = {0};
> +
> +INIT_OPT_KERN_SYM(mt_slots);
> +INIT_OPT_KERN_SYM(mt_pivots);
> +
> +INIT_OPT_KERN_STRUCT(maple_tree);
> +INIT_OPT_KERN_STRUCT(maple_node);
> +INIT_OPT_KERN_STRUCT_MEMBER(maple_tree, ma_root);
> +INIT_OPT_KERN_STRUCT_MEMBER(maple_node, ma64);
> +INIT_OPT_KERN_STRUCT_MEMBER(maple_node, mr64);
> +INIT_OPT_KERN_STRUCT_MEMBER(maple_node, slot);
> +INIT_OPT_KERN_STRUCT_MEMBER(maple_arange_64, pivot);
> +INIT_OPT_KERN_STRUCT_MEMBER(maple_arange_64, slot);
> +INIT_OPT_KERN_STRUCT_MEMBER(maple_range_64, pivot);
> +INIT_OPT_KERN_STRUCT_MEMBER(maple_range_64, slot);
> +
> +#define MEMBER_OFF(S, M) \
> +	GET_KERN_STRUCT_MEMBER_MOFF(S, M) / 8
> +
> +#define MAPLE_BUFSIZE			512
> +
> +enum {
> +	maple_dense_enum,
> +	maple_leaf_64_enum,
> +	maple_range_64_enum,
> +	maple_arange_64_enum,
> +};
> +
> +#define MAPLE_NODE_MASK			255UL
> +#define MAPLE_NODE_TYPE_MASK		0x0F
> +#define MAPLE_NODE_TYPE_SHIFT		0x03
> +#define XA_ZERO_ENTRY			xa_mk_internal(257)
> +
> +static unsigned long xa_mk_internal(unsigned long v)
> +{
> +	return (v << 2) | 2;
> +}
> +
> +static bool xa_is_internal(unsigned long entry)
> +{
> +	return (entry & 3) == 2;
> +}
> +
> +static bool xa_is_node(unsigned long entry)
> +{
> +	return xa_is_internal(entry) && entry > 4096;
> +}
> +
> +static bool xa_is_value(unsigned long entry)
> +{
> +	return entry & 1;
> +}
> +
> +static bool xa_is_zero(unsigned long entry)
> +{
> +	return entry == XA_ZERO_ENTRY;
> +}
> +
> +static unsigned long xa_to_internal(unsigned long entry)
> +{
> +	return entry >> 2;
> +}
> +
> +static unsigned long xa_to_value(unsigned long entry)
> +{
> +	return entry >> 1;
> +}
> +
> +static unsigned long mte_to_node(unsigned long entry)
> +{
> +        return entry & ~MAPLE_NODE_MASK;
> +}
> +
> +static unsigned long mte_node_type(unsigned long maple_enode_entry)
> +{
> +	return (maple_enode_entry >> MAPLE_NODE_TYPE_SHIFT) &
> +		MAPLE_NODE_TYPE_MASK;
> +}
> +
> +static unsigned long mt_slot(void **slots, unsigned char offset)
> +{
> +       return (unsigned long)slots[offset];
> +}
> +
> +static bool ma_is_leaf(unsigned long type)
> +{
> +	return type < maple_range_64_enum;
> +}
> +
> +static bool mte_is_leaf(unsigned long maple_enode_entry)
> +{
> +       return ma_is_leaf(mte_node_type(maple_enode_entry));
> +}
> +
> +static void mt_dump_entry(unsigned long entry, unsigned long min,
> +			unsigned long max, unsigned int depth,
> +			unsigned long **array_out, int *array_len,
> +			int *array_cap)
> +{
> +	if (entry == 0)
> +		return;
> +
> +	add_to_arr((void ***)array_out, array_len, array_cap, (void *)entry);
> +}
> +
> +static void mt_dump_node(unsigned long entry, unsigned long min,
> +			unsigned long max, unsigned int depth,
> +			unsigned long **array_out, int *array_len,
> +			int *array_cap);
> +
> +static void mt_dump_range64(unsigned long entry, unsigned long min,
> +			unsigned long max, unsigned int depth,
> +			unsigned long **array_out, int *array_len,
> +			int *array_cap)
> +{
> +	unsigned long maple_node_m_node = mte_to_node(entry);
> +	char node_buf[MAPLE_BUFSIZE];
> +	bool leaf = mte_is_leaf(entry);
> +	unsigned long first = min, last;
> +	int i;
> +	char *mr64_buf;
> +
> +	readmem(VADDR, maple_node_m_node, node_buf, GET_KERN_STRUCT_SSIZE(maple_node));
> +	mr64_buf = node_buf + MEMBER_OFF(maple_node, mr64);
> +
> +	for (i = 0; i < mt_slots[maple_range_64_enum]; i++) {
> +		last = max;
> +
> +		if (i < (mt_slots[maple_range_64_enum] - 1))
> +			last = ULONG(mr64_buf + MEMBER_OFF(maple_range_64, pivot) +
> +				     sizeof(ulong) * i);
> +
> +		else if (!VOID_PTR(mr64_buf + MEMBER_OFF(maple_range_64, slot) +
> +			  sizeof(void *) * i) &&
> +			 max != mt_max[mte_node_type(entry)])
> +			break;
> +		if (last == 0 && i > 0)
> +			break;
> +		if (leaf)
> +			mt_dump_entry(mt_slot((void **)(mr64_buf +
> +						      MEMBER_OFF(maple_range_64, slot)), i),
> +				first, last, depth + 1, array_out, array_len, array_cap);
> +		else if (VOID_PTR(mr64_buf + MEMBER_OFF(maple_range_64, slot) +
> +				  sizeof(void *) * i)) {
> +			mt_dump_node(mt_slot((void **)(mr64_buf +
> +						     MEMBER_OFF(maple_range_64, slot)), i),
> +				first, last, depth + 1, array_out, array_len, array_cap);
> +		}
> +
> +		if (last == max)
> +			break;
> +		if (last > max) {
> +			printf("node %p last (%lu) > max (%lu) at pivot %d!\n",
> +				mr64_buf, last, max, i);
> +			break;
> +		}
> +		first = last + 1;
> +	}
> +}
> +
> +static void mt_dump_arange64(unsigned long entry, unsigned long min,
> +			unsigned long max, unsigned int depth,
> +			unsigned long **array_out, int *array_len,
> +			int *array_cap)
> +{
> +	unsigned long maple_node_m_node = mte_to_node(entry);
> +	char node_buf[MAPLE_BUFSIZE];
> +	unsigned long first = min, last;
> +	int i;
> +	char *ma64_buf;
> +
> +	readmem(VADDR, maple_node_m_node, node_buf, GET_KERN_STRUCT_SSIZE(maple_node));
> +	ma64_buf = node_buf + MEMBER_OFF(maple_node, ma64);
> +
> +	for (i = 0; i < mt_slots[maple_arange_64_enum]; i++) {
> +		last = max;
> +
> +		if (i < (mt_slots[maple_arange_64_enum] - 1))
> +			last = ULONG(ma64_buf + MEMBER_OFF(maple_arange_64, pivot) +
> +				     sizeof(void *) * i);
> +		else if (!VOID_PTR(ma64_buf + MEMBER_OFF(maple_arange_64, slot) +
> +				   sizeof(void *) * i))
> +			break;
> +		if (last == 0 && i > 0)
> +			break;
> +
> +		if (ULONG(ma64_buf + MEMBER_OFF(maple_arange_64, slot) + sizeof(void *) * i))
> +			mt_dump_node(mt_slot((void **)(ma64_buf +
> +						      MEMBER_OFF(maple_arange_64, slot)), i),
> +				first, last, depth + 1, array_out, array_len, array_cap);
> +
> +		if (last == max)
> +			break;
> +		if (last > max) {
> +			printf("node %p last (%lu) > max (%lu) at pivot %d!\n",
> +				ma64_buf, last, max, i);
> +			break;
> +		}
> +		first = last + 1;
> +	}
> +}
> +
> +static void mt_dump_node(unsigned long entry, unsigned long min,
> +			unsigned long max, unsigned int depth,
> +			unsigned long **array_out, int *array_len,
> +			int *array_cap)
> +{
> +	unsigned long maple_node = mte_to_node(entry);
> +	unsigned long type = mte_node_type(entry);
> +	int i;
> +	char node_buf[MAPLE_BUFSIZE];
> +
> +	readmem(VADDR, maple_node, node_buf, GET_KERN_STRUCT_SSIZE(maple_node));
> +
> +	switch (type) {
> +	case maple_dense_enum:
> +		for (i = 0; i < mt_slots[maple_dense_enum]; i++) {
> +			if (min + i > max)
> +				printf("OUT OF RANGE: ");
> +			mt_dump_entry(mt_slot((void **)(node_buf + MEMBER_OFF(maple_node, slot)), i),
> +				min + i, min + i, depth, array_out, array_len, array_cap);
> +		}
> +		break;
> +	case maple_leaf_64_enum:
> +	case maple_range_64_enum:
> +		mt_dump_range64(entry, min, max, depth, array_out, array_len, array_cap);
> +		break;
> +	case maple_arange_64_enum:
> +		mt_dump_arange64(entry, min, max, depth, array_out, array_len, array_cap);
> +		break;
> +	default:
> +		printf(" UNKNOWN TYPE\n");
> +	}	
> +}
> +
> +unsigned long *mt_dump(unsigned long mt, int *array_len)
> +{
> +	char tree_buf[MAPLE_BUFSIZE];
> +	unsigned long entry;
> +	unsigned long *array_out = NULL;
> +	int array_cap = 0;
> +	*array_len = 0;
> +
> +	readmem(VADDR, mt, tree_buf, GET_KERN_STRUCT_SSIZE(maple_tree));
> +	entry = ULONG(tree_buf + MEMBER_OFF(maple_tree, ma_root));
> +
> +	if (xa_is_node(entry))
> +		mt_dump_node(entry, 0, mt_max[mte_node_type(entry)], 0,
> +				&array_out, array_len, &array_cap);
> +	else if (entry)
> +		mt_dump_entry(entry, 0, 0, 0, &array_out, array_len, &array_cap);
> +	else
> +		printf("(empty)\n");
> +
> +	return array_out;
> +}
> +
> +bool maple_init(void)
> +{
> +	unsigned long mt_slots_ptr;
> +	unsigned long mt_pivots_ptr;
> +
> +	if (!KERN_SYM_EXIST(mt_slots) ||
> +	    !KERN_SYM_EXIST(mt_pivots) ||
> +	    !KERN_STRUCT_EXIST(maple_tree) ||
> +	    !KERN_STRUCT_EXIST(maple_node) ||
> +	    !KERN_STRUCT_MEMBER_EXIST(maple_tree, ma_root) ||
> +	    !KERN_STRUCT_MEMBER_EXIST(maple_node, ma64) ||
> +	    !KERN_STRUCT_MEMBER_EXIST(maple_node, mr64) ||
> +	    !KERN_STRUCT_MEMBER_EXIST(maple_node, slot) ||
> +	    !KERN_STRUCT_MEMBER_EXIST(maple_arange_64, pivot) ||
> +	    !KERN_STRUCT_MEMBER_EXIST(maple_arange_64, slot) ||
> +	    !KERN_STRUCT_MEMBER_EXIST(maple_range_64, pivot) ||
> +	    !KERN_STRUCT_MEMBER_EXIST(maple_range_64, slot)) {
> +		printf("%s: Missing required maple tree syms/types\n",
> +			__func__);
> +		return false;
> +	}
> +
> +	mt_slots_ptr = GET_KERN_SYM(mt_slots);
> +	mt_pivots_ptr = GET_KERN_SYM(mt_pivots);
> +
> +	if (GET_KERN_STRUCT_SSIZE(maple_tree) > MAPLE_BUFSIZE ||
> +	    GET_KERN_STRUCT_SSIZE(maple_node) > MAPLE_BUFSIZE) {
> +		printf("%s: MAPLE_BUFSIZE should be larger than maple_node/tree struct\n",
> +			__func__);
> +		return false;
> +	}
> +
> +	readmem(VADDR, mt_slots_ptr, mt_slots, sizeof(mt_slots));
> +	readmem(VADDR, mt_pivots_ptr, mt_pivots, sizeof(mt_pivots));
> +
> +	mt_max[maple_dense_enum]           = mt_slots[maple_dense_enum];
> +	mt_max[maple_leaf_64_enum]         = ULONG_MAX;
> +	mt_max[maple_range_64_enum]        = ULONG_MAX;
> +	mt_max[maple_arange_64_enum]       = ULONG_MAX;
> +
> +	return true;
> +}
> \ No newline at end of file
> diff --git a/extensions/maple_tree.h b/extensions/maple_tree.h
> new file mode 100644
> index 0000000..c96624c
> --- /dev/null
> +++ b/extensions/maple_tree.h
> @@ -0,0 +1,6 @@
> +#ifndef _MAPLE_TREE_H
> +#define _MAPLE_TREE_H
> +#include <stdbool.h>
> +unsigned long *mt_dump(unsigned long mt, int *array_len);
> +bool maple_init(void);
> +#endif /* _MAPLE_TREE_H */
> \ No newline at end of file
> -- 
> 2.47.0