rt/snapraid
1
0
Fork 0
mirror of https://github.com/amadvance/snapraid.git synced 2026-10-26 10:51:54 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Defer deallocation of modified files to phase 2 to prevent copy-detection races

Browse source
This commit is contained in:
Andrea Mazzoleni 2026-07-16 23:22:15 +02:00
parent 821756371c
commit 4ebdfe8c8f
3 changed files with 92 additions and 34 deletions
Download patch file Download diff file Expand all files Collapse all files

1
cmdline/elem.c
View file

@ -437,6 +437,7 @@ struct snapraid_file* file_dup(struct snapraid_file* copy)
file->inode = copy->inode;
file->physical = copy->physical;
file->flag = copy->flag;
file->shared_flag = 0;
file->blockvec = nalloc_nofail(file->blockmax, block_sizeof());
for (i = 0; i < file->blockmax; ++i) {

32
cmdline/elem.h
View file

@ -245,11 +245,26 @@ struct snapraid_block {
#define FILE_IS_JUNCTION 0x8000 /**< If it's a junction for Windows. Not yet supported. */
#define FILE_IS_LINK_MASK 0xF000 /**< Mask for link type. */
/**
* Flag to indicate the old version of a modified file during scanning.
*/
#define FILE_IS_MODIFIED_OLD 0x10000
/**
* Flag to indicate the new version of a modified/reallocated file during scanning.
*/
#define FILE_IS_MODIFIED_NEW 0x20000
/**
* Flag to indicate the old version of a kept file being reallocated during scanning.
*/
#define FILE_IS_REALLOC_OLD 0x40000
/**
* Flags from this bit are shared between multiple threads
* and goes in the shared_flags instead of flags
*/
#define FILE_FLAGS_SHARED_BIT 16
#define FILE_FLAGS_SHARED_BIT 24
/**
* During scan if the file is found missing and relocated to another place
@ -259,7 +274,7 @@ struct snapraid_block {
* This specific bit is written using protection from the stamp_mutex during
* the multithread scan.
*/
#define FILE_IS_RELOCATED 0x10000
#define FILE_IS_RELOCATED 0x1000000
/**
* File.
@ -273,8 +288,8 @@ struct snapraid_file {
block_off_t blockmax; /**< Number of blocks. */
char* sub; /**< Sub path of the file. Without the disk dir. The disk is implicit. */
int mtime_nsec; /**< Modification time nanoseconds. In the range 0 <= x < 1,000,000,000, or STAT_NSEC_INVALID if not present. */
uint16_t flag; /**< FILE_IS_* flags. */
uint16_t shared_flag; /**< FILE_IS_RELOCATED flag. Keep it separated as it's accessed by multiple threads */
uint32_t flag; /**< FILE_IS_* flags. */
uint32_t shared_flag; /**< FILE_IS_RELOCATED flag. Keep it separated as it's accessed by multiple threads */
/* nodes for data structures */
tommy_node nodelist;
@ -800,8 +815,7 @@ static inline int block_has_file_and_valid_parity(const struct snapraid_block* b
static inline int file_flag_has(const struct snapraid_file* file, unsigned mask)
{
if (mask >= (1 << FILE_FLAGS_SHARED_BIT)) {
mask >>= FILE_FLAGS_SHARED_BIT;
if (mask >= (1U << FILE_FLAGS_SHARED_BIT)) {
return (file->shared_flag & mask) == mask;
} else {
return (file->flag & mask) == mask;
@ -810,8 +824,7 @@ static inline int file_flag_has(const struct snapraid_file* file, unsigned mask)
static inline void file_flag_set(struct snapraid_file* file, unsigned mask)
{
if (mask >= (1 << FILE_FLAGS_SHARED_BIT)) {
mask >>= FILE_FLAGS_SHARED_BIT;
if (mask >= (1U << FILE_FLAGS_SHARED_BIT)) {
file->shared_flag |= mask;
} else {
file->flag |= mask;
@ -820,8 +833,7 @@ static inline void file_flag_set(struct snapraid_file* file, unsigned mask)
static inline void file_flag_clear(struct snapraid_file* file, unsigned mask)
{
if (mask >= (1 << FILE_FLAGS_SHARED_BIT)) {
mask >>= FILE_FLAGS_SHARED_BIT;
if (mask >= (1U << FILE_FLAGS_SHARED_BIT)) {
file->shared_flag &= ~mask;
} else {
file->flag &= ~mask;

93
cmdline/scan.c
View file

@ -417,7 +417,7 @@ static void scan_file_delayed_allocate(struct snapraid_scan* scan, struct snapra
}
}
/* insert in the delayed list */
/* insert in the delayed allocation list */
tommy_list_insert_tail(&scan->file_insert_list, &file->nodelist, file);
}
@ -640,9 +640,6 @@ static void scan_file_insert(struct snapraid_scan* scan, struct snapraid_file* f
tommy_hashdyn_insert(&disk->pathset, &file->pathset, file, file_path_hash(file->sub));
tommy_hashdyn_insert(&disk->stampset, &file->stampset, file, file_stamp_hash(file->size, file->mtime_sec, file->mtime_nsec));
stamp_unlock(disk);
/* delayed allocation of the parity */
scan_file_delayed_allocate(scan, file);
}
/**
@ -677,7 +674,15 @@ static void scan_file_remove(struct snapraid_scan* scan, struct snapraid_file* f
tommy_hashdyn_remove_existing(&disk->stampset, &file->stampset);
stamp_unlock(disk);
/* deallocate the file from the parity */
/*
* Deallocate the file from the parity.
*
* This is safe to run unlocked because:
* 1. Modified files deallocations are deferred to Phase 2 (mono-threaded phase).
* 2. Invalid parity files deallocated in Phase 1 are never selected by
* file_is_full_hashed_and_stable() during copy-detection (since they lack
* valid parity blocks).
*/
scan_file_deallocate(scan, file);
}
@ -698,13 +703,15 @@ static void scan_file_keep(struct snapraid_scan* scan, struct snapraid_file* fil
/* if the file is full invalid, schedule a reinsert at later stage */
if (file_is_full_invalid_parity_and_stable(scan->state, disk, file)) {
struct snapraid_file* copy = file_dup(file);
file_flag_set(copy, FILE_IS_MODIFIED_NEW);
/* remove the file */
scan_file_remove(scan, file, 0);
/* insert in the delayed allocation list */
scan_file_delayed_allocate(scan, copy);
/* reinsert the copy in the delayed list */
scan_file_insert(scan, copy);
/* mark the file to be removed on Phase 2 */
file_flag_set(file, FILE_IS_REALLOC_OLD);
}
}
@ -723,6 +730,7 @@ static void scan_file(struct snapraid_scan* scan, int is_diff, const char* sub,
int64_t file_already_present_mtime_sec;
int file_already_present_mtime_nsec;
int is_file_reported;
int is_file_modified;
/*
* If the disk has persistent inodes and UUID, try a search on the past inodes,
* to detect moved files.
@ -912,6 +920,7 @@ static void scan_file(struct snapraid_scan* scan, int is_diff, const char* sub,
/* initialize for later overwrite */
is_file_reported = 0;
is_file_modified = 0;
is_original_file_size_different_than_zero = 0;
/* then try finding it by name */
@ -1034,8 +1043,11 @@ static void scan_file(struct snapraid_scan* scan, int is_diff, const char* sub,
/* keep track if the original file was not of zero size */
is_original_file_size_different_than_zero = file->size != 0;
/* remove it, and continue to insert it again */
scan_file_remove(scan, file, 1);
/* mark it as to be removed in Phase 2 */
file_flag_set(file, FILE_IS_MODIFIED_OLD);
/* flag that we are inserting the modified version so we skip Phase 1 insertion sets */
is_file_modified = 1;
/* and continue to insert it again */
} else {
@ -1114,7 +1126,14 @@ static void scan_file(struct snapraid_scan* scan, int is_diff, const char* sub,
}
stamp_unlock(other_disk);
/* if found, and it's a fully hashed file */
/*
* If found, check stability and copy the hash.
*
* This is safe to execute unlocked because:
* 1. Modified files deallocations are deferred to Phase 2 (mono-threaded phase).
* 2. file_is_full_invalid_parity_and_stable() only accepts files with complete
* invalid parity, preventing them from being selected by file_is_full_hashed_and_stable().
*/
if (other_file && file_is_full_hashed_and_stable(scan->state, other_disk, other_file)) {
char path_other[PATH_MAX];
struct stat other_st;
@ -1179,8 +1198,14 @@ static void scan_file(struct snapraid_scan* scan, int is_diff, const char* sub,
}
}
/* insert the file in the delayed list */
scan_file_insert(scan, file);
if (!is_file_modified) { /* modified files are re-inserted in Phase 3 */
scan_file_insert(scan, file);
} else {
file_flag_set(file, FILE_IS_MODIFIED_NEW);
}
/* insert the file in the delayed allocation list */
scan_file_delayed_allocate(scan, file);
}
/**
@ -1759,7 +1784,20 @@ static int state_diffscan(struct snapraid_state* state, int is_diff)
tommy_list_insert_tail(&scanlist, &scan->node, scan);
}
/* first scan all the directory and find new and deleted files */
/*
* We split the search in three phases:
* Phase 1: Parallel scanning of directories, finding new and modified files (without deletions/deallocations).
* Phase 2: Serialized removals (deleted files and old versions of modified files) to free up parity space.
* Phase 3: Serialized insertions and allocations of new files.
*
* We must start Phase 2 (deletions) only when all disks have finished Phase 1 (scanning),
* to ensure that copy/relocation detection on any disk can search the stampset of other disks
* before their old files are deleted/deallocated.
*/
/*
* Phase 1: Parallel scanning of directories
*/
for (i = scanlist; i != 0; i = i->next) {
struct snapraid_scan* scan = i->data;
#if HAVE_THREAD
@ -1784,13 +1822,6 @@ static int state_diffscan(struct snapraid_state* state, int is_diff)
}
#endif
/*
* We split the search in two phases because to detect files
* moved from one disk to another we have to start deletion
* only when all disks have all the new files found
*/
/* now process all the new and deleted files */
for (i = scanlist; i != 0; i = i->next) {
struct snapraid_scan* scan = i->data;
struct snapraid_disk* disk = scan->disk;
@ -1799,6 +1830,10 @@ static int state_diffscan(struct snapraid_state* state, int is_diff)
uint64_t phy_last;
struct snapraid_file* phy_file_last;
/*
* Phase 2: Removals (deleted files and old versions of modified files)
*/
/* check for removed files */
node = disk->filelist;
while (node) {
@ -1807,8 +1842,11 @@ static int state_diffscan(struct snapraid_state* state, int is_diff)
/* next node */
node = node->next;
/* remove if not present */
if (!file_flag_has(file, FILE_IS_PRESENT)) {
if (file_flag_has(file, FILE_IS_REALLOC_OLD)) {
scan_file_remove(scan, file, 0);
} else if (file_flag_has(file, FILE_IS_MODIFIED_OLD)) {
scan_file_remove(scan, file, 1);
} else if (!file_flag_has(file, FILE_IS_PRESENT)) {
/* here we are in mono-thread context, no need to use the stamp_lock() to read FILE_IS_RELOCATED */
if (!file_flag_has(file, FILE_IS_RELOCATED)) {
++scan->count_remove;
@ -1859,6 +1897,8 @@ static int state_diffscan(struct snapraid_state* state, int is_diff)
}
/*
* Phase 3: Insertions (new files and new versions of modified files)
*
* Sort the files before inserting them
* we use a stable sort to ensure that if the reported physical offset/inode
* are always 0, we keep at least the directory order
@ -1912,6 +1952,11 @@ static int state_diffscan(struct snapraid_state* state, int is_diff)
/* next node */
node = node->next;
/* if this is a modified file new version, insert it into sets now */
if (file_flag_has(file, FILE_IS_MODIFIED_NEW)) {
scan_file_insert(scan, file);
}
/* insert in the parity */
scan_file_allocate(scan, file);
}