52th

arch/arm64/kernel/sys.c

@@ -18,6 +18,43 @@
 #include <asm/cpufeature.h>
 #include <asm/syscall.h>
 
+/*
+ * IAMROOT, 2022.05.21:
+ * - arm64 sys_mmap
+ *
+ * ex) user에서 malloc을 사용했을때 strace 사용결과
+ * sh) strace ./test
+ * == strace
+ * mmap(NULL, 1000001536, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7fb4af7ee000
+ *
+ * == smaps log(heap 측 부분)
+ * 7fb4af7ee000-7fb4eb19b000 rw-p 00000000 00:00 0 
+ * Size:             976564 kB
+ * KernelPageSize:        4 kB
+ * MMUPageSize:           4 kB
+ * Rss:              976564 kB
+ * Pss:              976564 kB
+ * Shared_Clean:          0 kB
+ * Shared_Dirty:          0 kB
+ * Private_Clean:         0 kB
+ * Private_Dirty:    976564 kB
+ * Referenced:       976564 kB
+ * Anonymous:        976564 kB
+ * LazyFree:              0 kB
+ * AnonHugePages:         0 kB
+ * ShmemPmdMapped:        0 kB
+ * FilePmdMapped:         0 kB
+ * Shared_Hugetlb:        0 kB
+ * Private_Hugetlb:       0 kB
+ * Swap:                  0 kB
+ * SwapPss:               0 kB
+ * Locked:                0 kB
+ * THPeligible:    0
+ * VmFlags: rd wr mr(may read) mw(may write) me(may execute) ac sd 
+ *
+ * read, write 권한이 있는 vma라는것을 알수있다.
+ * malloc을 호출하면 항상 위와같은 prot, flag를 사용한다.
+ */
 SYSCALL_DEFINE6(mmap, unsigned long, addr, unsigned long, len,
 		unsigned long, prot, unsigned long, flags,
 		unsigned long, fd, unsigned long, off)

include/asm-generic/tlb.h

@@ -498,6 +498,11 @@ static inline void tlb_start_vma(struct mmu_gather *tlb, struct vm_area_struct *
 #endif
 
 #ifndef tlb_end_vma
+
+/*
+ * IAMROOT, 2022.05.21:
+ * - tlb flush 
+ */
 static inline void tlb_end_vma(struct mmu_gather *tlb, struct vm_area_struct *vma)
 {
 	if (tlb->fullmm)

include/linux/mm.h

@@ -2842,6 +2842,11 @@ extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned lon
  * Returns: The first VMA within the provided range, %NULL otherwise.  Assumes
  * start_addr < end_addr.
  */
+
+/*
+ * IAMROOT, 2022.05.21:
+ * - 겹치는 vma를 찾아온다.
+ */
 static inline
 struct vm_area_struct *find_vma_intersection(struct mm_struct *mm,
 					     unsigned long start_addr,
@@ -2872,6 +2877,10 @@ struct vm_area_struct *vma_lookup(struct mm_struct *mm, unsigned long addr)
 	return vma;
 }
 
+/*
+ * IAMROOT, 2022.05.21:
+ * - @vma가 아래로 내려가는 stack인 경우 보정.
+ */
 static inline unsigned long vm_start_gap(struct vm_area_struct *vma)
 {
 	unsigned long vm_start = vma->vm_start;
@@ -2884,6 +2893,10 @@ static inline unsigned long vm_start_gap(struct vm_area_struct *vma)
 	return vm_start;
 }
 
+/*
+ * IAMROOT, 2022.05.21:
+ * - @vma가 위로 올라가는 stack인 경우 보정.
+ */
 static inline unsigned long vm_end_gap(struct vm_area_struct *vma)
 {
 	unsigned long vm_end = vma->vm_end;

include/linux/mm_types.h

@@ -434,6 +434,13 @@ struct vm_area_struct {
 	 * VMAs below us in the VMA rbtree and its ->vm_prev. This helps
 	 * get_unmapped_area find a free area of the right size.
 	 */
+/*
+ * IAMROOT, 2022.05.21:
+ * - 이 VMA의 왼쪽에 있는 최대 사용 가능한 메모리 간격(바이트)입니다.
+ *   이 VMA와 vma->vm_prev 사이 또는 VMArbtree에서 아래 VMA 중 하나와
+ *   해당 ->vm_prev 사이입니다. 이렇게 하면 get_unmapped_area가 올바른
+ *   크기의 빈 영역을 찾을 수 있습니다. 
+ */
 	unsigned long rb_subtree_gap;
 
 	/* Second cache line starts here. */
@@ -594,6 +601,11 @@ struct mm_struct {
 		unsigned long total_vm;	   /* Total pages mapped */
 		unsigned long locked_vm;   /* Pages that have PG_mlocked set */
 		atomic64_t    pinned_vm;   /* Refcount permanently increased */
+/*
+ * IAMROOT, 2022.05.21:
+ * - data공간이라 하면 주석과 같이 not stack, not shared를 의미한다.
+ *   (is_data_mapping() 참고)
+ */
 		unsigned long data_vm;	   /* VM_WRITE & ~VM_SHARED & ~VM_STACK */
 		unsigned long exec_vm;	   /* VM_EXEC & ~VM_WRITE & ~VM_STACK */
 		unsigned long stack_vm;	   /* VM_STACK */

include/linux/mman.h

@@ -130,6 +130,10 @@ static inline bool arch_validate_flags(unsigned long flags)
  * but this version is faster.
  * ("bit1" and "bit2" must be single bits)
  */
+/*
+ * IAMROOT, 2022.05.21:
+ * - bit1이 있으면 bit2 return 이라는뜻.
+ */
 #define _calc_vm_trans(x, bit1, bit2) \
   ((!(bit1) || !(bit2)) ? 0 : \
   ((bit1) <= (bit2) ? ((x) & (bit1)) * ((bit2) / (bit1)) \
@@ -138,6 +142,16 @@ static inline bool arch_validate_flags(unsigned long flags)
 /*
  * Combine the mmap "prot" argument into "vm_flags" used internally.
  */
+/*
+ * IAMROOT, 2022.05.21:
+ * - prot flag를 vm flag로 변환하고 arch 에 따른 flag 추가가 있으면
+ *   추가 한다.
+ * - PROT_READ -> VM_READ
+ *   PROT_WRITE -> VM_WRITE
+ *   PROT_EXEC -> VM_EXEC
+ *   PROT_BTI  -> VM_ARM64_BTI
+ *   PROT_MTE  -> VM_MTE
+ */
 static inline unsigned long
 calc_vm_prot_bits(unsigned long prot, unsigned long pkey)
 {
@@ -150,6 +164,14 @@ calc_vm_prot_bits(unsigned long prot, unsigned long pkey)
 /*
  * Combine the mmap "flags" argument into "vm_flags" used internally.
  */
+/*
+ * IAMROOT, 2022.05.21:
+ * - map flags를 vm flags로 변환한다.
+ * - MAP_GROWSDOWN -> VM_GROWSDOWN
+ *   MAP_LOCKED -> VM_LOCKED
+ *   MAP_SYNC -> VM_SYNC
+ *   MAP_ANONYMOUS -> VM_MTE_ALLOWED
+ */
 static inline unsigned long
 calc_vm_flag_bits(unsigned long flags)
 {

include/linux/sched/signal.h

@@ -751,6 +751,10 @@ static inline unsigned long task_rlimit_max(const struct task_struct *task,
 	return READ_ONCE(task->signal->rlim[limit].rlim_max);
 }
 
+/*
+ * IAMROOT, 2022.05.21:
+ * - 해당 task가 @limit에 해당하는 제한 값을 읽어온다. 
+ */
 static inline unsigned long rlimit(unsigned int limit)
 {
 	return task_rlimit(current, limit);

include/uapi/asm-generic/mman-common.h

@@ -19,12 +19,27 @@
 
 /* 0x01 - 0x03 are defined in linux/mman.h */
 #define MAP_TYPE	0x0f		/* Mask for type of mapping */
+
+/*
+ * IAMROOT, 2022.05.21:
+ * - 가상공간의 특정영역에 고정시키고 싶을때 사용한다.
+ */
 #define MAP_FIXED	0x10		/* Interpret addr exactly */
 #define MAP_ANONYMOUS	0x20		/* don't use a file */
 
 /* 0x0100 - 0x4000 flags are defined in asm-generic/mman.h */
+
+/*
+ * IAMROOT, 2022.05.21:
+ * - mapping을 바로 하고싶을때 사용한다.
+ */
 #define MAP_POPULATE		0x008000	/* populate (prefault) pagetables */
 #define MAP_NONBLOCK		0x010000	/* do not block on IO */
+
+/*
+ * IAMROOT, 2022.05.21:
+ * - user stack만들때 brk를 호출하거나 MAP_STACK을 사용해서 sys_mmap을 호출에서 사용하게 될것이다.
+ */
 #define MAP_STACK		0x020000	/* give out an address that is best suited for process/thread stacks */
 #define MAP_HUGETLB		0x040000	/* create a huge page mapping */
 #define MAP_SYNC		0x080000 /* perform synchronous page faults for the mapping */

include/uapi/asm-generic/resource.h

@@ -13,6 +13,10 @@
  *   then it defines them prior including asm-generic/resource.h. )
  */
 
+/*
+ * IAMROOT, 2022.05.21:
+ * - resource 제한에 대한것들.
+ */
 #define RLIMIT_CPU		0	/* CPU time in sec */
 #define RLIMIT_FSIZE		1	/* Maximum filesize */
 #define RLIMIT_DATA		2	/* max data size */

mm/internal.h

@@ -467,6 +467,11 @@ static inline bool is_stack_mapping(vm_flags_t flags)
 /*
  * Data area - private, writable, not stack
  */
+/*
+ * IAMROOT, 2022.05.21:
+ * stack을 제외한 data 공간.
+ * Data area - private, writable, not stack
+ */
 static inline bool is_data_mapping(vm_flags_t flags)
 {
 	return (flags & (VM_WRITE | VM_SHARED | VM_STACK)) == VM_WRITE;

mm/memory.c

@@ -1539,6 +1539,10 @@ static inline unsigned long zap_p4d_range(struct mmu_gather *tlb,
 	return addr;
 }
 
+/*
+ * IAMROOT, 2022.05.21:
+ * - 해당 영역의 정규 mapping을 unmap
+ */
 void unmap_page_range(struct mmu_gather *tlb,
 			     struct vm_area_struct *vma,
 			     unsigned long addr, unsigned long end,
@@ -1556,10 +1560,17 @@ void unmap_page_range(struct mmu_gather *tlb,
 			continue;
 		next = zap_p4d_range(tlb, vma, pgd, addr, next, details);
 	} while (pgd++, addr = next, addr != end);
+/*
+ * IAMROOT, 2022.05.21:
+ * - @vma에 대한 tlb flush
+ */
 	tlb_end_vma(tlb, vma);
 }
 
-
+/*
+ * IAMROOT, 2022.05.21:
+ * - @vma unmmap
+ */
 static void unmap_single_vma(struct mmu_gather *tlb,
 		struct vm_area_struct *vma, unsigned long start_addr,
 		unsigned long end_addr,
@@ -1621,6 +1632,33 @@ static void unmap_single_vma(struct mmu_gather *tlb,
  * ensure that any thus-far unmapped pages are flushed before unmap_vmas()
  * drops the lock and schedules.
  */
+
+/*
+ * IAMROOT, 2022.05.21:
+ *
+ * ^ ---- vm_end
+ * |                    <- vma4
+ * |                --- end
+ * |                 ^  <- vma3->next = vma4
+ * v ---- vm_start   |
+ *                   |
+ * ...               ..
+ *                    
+ * ^ ---- vm_end     |
+ * |                 |
+ * |                 | <- vma2->next = vma3
+ * v ---- vm_start   |
+ *                   |
+ * ...              ..
+ *                   |
+ * ^ ---- vm_end     |
+ * |                 v  <- vma->next = vma2
+ * |                --- start
+ * |                 <-- prev
+ * v ---- vm_start
+ * 
+ * vma부터 ~ vma4까지  unmap될것이다.(unmap_vmas())
+ */
 void unmap_vmas(struct mmu_gather *tlb,
 		struct vm_area_struct *vma, unsigned long start_addr,
 		unsigned long end_addr)
@@ -3717,6 +3755,10 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
  * but allow concurrent faults), and pte mapped but not yet locked.
  * We return with mmap_lock still held, but pte unmapped and unlocked.
  */
+/*
+ * IAMROOT, 2022.05.21:
+ * - anon page fault시 이 함수로 진입하게 될것이다.
+ */
 static vm_fault_t do_anonymous_page(struct vm_fault *vmf)
 {
 	struct vm_area_struct *vma = vmf->vma;

mm/mempolicy.c

@@ -1840,6 +1840,10 @@ nodemask_t *policy_nodemask(gfp_t gfp, struct mempolicy *policy)
  * policy_node() is always coupled with policy_nodemask(), which
  * secures the nodemask limit for 'bind' and 'prefer-many' policy.
  */
+/*
+ * IAMROOT, 2022.05.21:
+ * - @policy->mode 가 MPOL_PREFERRED인 경우 first node를 return한다.
+ */
 static int policy_node(gfp_t gfp, struct mempolicy *policy, int nd)
 {
 	if (policy->mode == MPOL_PREFERRED) {
@@ -2193,6 +2197,14 @@ static struct page *alloc_pages_preferred_many(gfp_t gfp, unsigned int order,
  *
  * Return: The page on success or NULL if allocation fails.
  */
+/*
+ * IAMROOT, 2022.05.21:
+ * - numa Policy는 주로 app을 위한 정책이다. policy에 따라 alloc_pages를 호출한다.
+ *
+ * ---
+ * - kernel같은 경우은 local이 preferred 인 개념이되고(나머지는 cost순),
+ *   app같은 경우엔 사용자가 설정한 node policy에 따르게 된다.
+ */
 struct page *alloc_pages_vma(gfp_t gfp, int order, struct vm_area_struct *vma,
 		unsigned long addr, int node, bool hugepage)
 {
@@ -2218,6 +2230,10 @@ struct page *alloc_pages_vma(gfp_t gfp, int order, struct vm_area_struct *vma,
 		goto out;
 	}
 
+/*
+ * IAMROOT, 2022.05.21:
+ * - PASS
+ */
 	if (unlikely(IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE) && hugepage)) {
 		int hpage_node = node;