{- Copying files.
-
- - Copyright 2011-2022 Joey Hess <id@joeyh.name>
+ - Copyright 2011-2025 Joey Hess <id@joeyh.name>
-
- Licensed under the GNU AGPL version 3 or higher.
-}
module Annex.CopyFile where
import Annex.Common
+import qualified Annex
import Utility.Metered
import Utility.CopyFile
import Utility.FileMode
data CopyMethod = CopiedCoW | Copied
+-- Should cp be allowed to copy the file with --reflink=auto?
+--
+-- The benefit is that this lets it use the copy_file_range
+-- syscall, which is not used with --reflink=always. The drawback is that
+-- the IncrementalVerifier is not updated, so verification, if it is done,
+-- will need to re-read the whole content of the file. And, interrupted
+-- copies are not resumed but are restarted from the beginning.
+--
+-- Using this will result in CopiedCow being returned even in cases
+-- where cp fell back to a slow copy.
+newtype FastCopy = FastCopy Bool
+
+getFastCopy :: RemoteGitConfig -> Annex FastCopy
+getFastCopy gc = case remoteAnnexFastCopy gc of
+ False -> FastCopy . annexFastCopy <$> Annex.getGitConfig
+ True -> return (FastCopy True)
+
{- Copies from src to dest, updating a meter. Preserves mode and mtime.
- Uses copy-on-write if it is supported. If the the destination already
- exists, an interrupted copy will resume where it left off.
- (eg when isStableKey is false), and doing this avoids getting a
- corrupted file in such cases.
-}
-fileCopier :: CopyCoWTried -> OsPath -> OsPath -> MeterUpdate -> Maybe IncrementalVerifier -> IO CopyMethod
+fileCopier :: CopyCoWTried -> FastCopy -> OsPath -> OsPath -> MeterUpdate -> Maybe IncrementalVerifier -> IO CopyMethod
+fileCopier copycowtried (FastCopy True) src dest meterupdate iv = do
+ ok <- watchFileSize dest meterupdate $ const $
+ copyFileExternal CopyTimeStamps src dest
+ if ok
+ then do
+ maybe noop unableIncrementalVerifier iv
+ return CopiedCoW
+ else fileCopier copycowtried (FastCopy False) src dest meterupdate iv
#ifdef mingw32_HOST_OS
-fileCopier _ src dest meterupdate iv = docopy
+fileCopier _ _ src dest meterupdate iv =
+ fileCopier' src dest meterupdate iv
#else
-fileCopier copycowtried src dest meterupdate iv =
+fileCopier copycowtried _ src dest meterupdate iv =
ifM (tryCopyCoW copycowtried src dest meterupdate)
( do
maybe noop unableIncrementalVerifier iv
return CopiedCoW
- , docopy
+ , fileCopier' src dest meterupdate iv
)
#endif
- where
- docopy = do
- -- The file might have had the write bit removed,
- -- so make sure we can write to it.
- void $ tryIO $ allowWrite dest
- F.withBinaryFile src ReadMode $ \hsrc ->
- fileContentCopier hsrc dest meterupdate iv
+fileCopier' :: OsPath -> OsPath -> MeterUpdate -> Maybe IncrementalVerifier -> IO CopyMethod
+fileCopier' src dest meterupdate iv = do
+ -- The file might have had the write bit removed,
+ -- so make sure we can write to it.
+ void $ tryIO $ allowWrite dest
+
+ F.withBinaryFile src ReadMode $ \hsrc ->
+ fileContentCopier hsrc dest meterupdate iv
- -- Copy src mode and mtime.
- mode <- fileMode <$> R.getFileStatus (fromOsPath src)
- mtime <- utcTimeToPOSIXSeconds <$> getModificationTime src
- let dest' = fromOsPath dest
- R.setFileMode dest' mode
- touch dest' mtime False
+ -- Copy src mode and mtime.
+ mode <- fileMode <$> R.getFileStatus (fromOsPath src)
+ mtime <- utcTimeToPOSIXSeconds <$> getModificationTime src
+ let dest' = fromOsPath dest
+ R.setFileMode dest' mode
+ touch dest' mtime False
- return Copied
+ return Copied
{- Copies content from a handle to a destination file. Does not
- use copy-on-write, and does not copy file mode and mtime.
+ - Updates the IncementalVerifier with the content it copies.
-}
fileContentCopier :: Handle -> OsPath -> MeterUpdate -> Maybe IncrementalVerifier -> IO ()
fileContentCopier hsrc dest meterupdate iv =
* map: Improve display of remote names.
* Windows: Fix duplicate file bug that could occur when files were
supposed to be moved across devices.
+ * Added annex.fastcopy and remote.name.annex-fastcopy config setting.
+ When set, this allows the copy_file_range syscall to be used, which
+ can eg allow for server-side copies on NFS. (For fastest copying,
+ also disable annex.verify or remote.name.annex-verify.)
-- Joey Hess <id@joeyh.name> Thu, 22 May 2025 12:43:38 -0400
cst <- remoteCost gc c cheapRemoteCost
let chunkconfig = getChunkConfig c
cow <- liftIO newCopyCoWTried
+ fastcopy <- getFastCopy gc
let ii = IgnoreInodes $ fromMaybe True $
getRemoteConfigValue ignoreinodesField c
return $ Just $ specialRemote c
- (storeKeyM dir chunkconfig cow)
- (retrieveKeyFileM dir chunkconfig cow)
+ (storeKeyM dir chunkconfig cow fastcopy)
+ (retrieveKeyFileM dir chunkconfig cow fastcopy)
(removeKeyM dir)
(checkPresentM dir chunkconfig)
Remote
, checkPresent = checkPresentDummy
, checkPresentCheap = True
, exportActions = ExportActions
- { storeExport = storeExportM dir cow
- , retrieveExport = retrieveExportM dir cow
+ { storeExport = storeExportM dir cow fastcopy
+ , retrieveExport = retrieveExportM dir cow fastcopy
, removeExport = removeExportM dir
, checkPresentExport = checkPresentExportM dir
-- Not needed because removeExportLocation
{ listImportableContents = listImportableContentsM ii dir
, importKey = Just (importKeyM ii dir)
, retrieveExportWithContentIdentifier = retrieveExportWithContentIdentifierM ii dir cow
- , storeExportWithContentIdentifier = storeExportWithContentIdentifierM ii dir cow
+ , storeExportWithContentIdentifier = storeExportWithContentIdentifierM ii dir cow fastcopy
, removeExportWithContentIdentifier = removeExportWithContentIdentifierM ii dir
-- Not needed because removeExportWithContentIdentifier
-- auto-removes empty directories.
{- Check if there is enough free disk space in the remote's directory to
- store the key. Note that the unencrypted key size is checked. -}
-storeKeyM :: OsPath -> ChunkConfig -> CopyCoWTried -> Storer
-storeKeyM d chunkconfig cow k c m =
+storeKeyM :: OsPath -> ChunkConfig -> CopyCoWTried -> FastCopy -> Storer
+storeKeyM d chunkconfig cow fastcopy k c m =
ifM (checkDiskSpaceDirectory d k)
( do
void $ liftIO $ tryIO $ createDirectoryUnder [d] tmpdir
in byteStorer go k c m
NoChunks ->
let go _k src p = liftIO $ do
- void $ fileCopier cow src tmpf p Nothing
+ void $ fileCopier cow fastcopy src tmpf p Nothing
finalizeStoreGeneric d tmpdir destdir
in fileStorer go k c m
_ ->
mapM_ preventWrite =<< dirContents dest
preventWrite dest
-retrieveKeyFileM :: OsPath -> ChunkConfig -> CopyCoWTried -> Retriever
-retrieveKeyFileM d (LegacyChunks _) _ = Legacy.retrieve locations' d
-retrieveKeyFileM d NoChunks cow = fileRetriever' $ \dest k p iv -> do
+retrieveKeyFileM :: OsPath -> ChunkConfig -> CopyCoWTried -> FastCopy -> Retriever
+retrieveKeyFileM d (LegacyChunks _) _ _ = Legacy.retrieve locations' d
+retrieveKeyFileM d NoChunks cow fastcopy = fileRetriever' $ \dest k p iv -> do
src <- liftIO $ getLocation d k
- void $ liftIO $ fileCopier cow src dest p iv
-retrieveKeyFileM d _ _ = byteRetriever $ \k sink ->
+ void $ liftIO $ fileCopier cow fastcopy src dest p iv
+retrieveKeyFileM d _ _ _ = byteRetriever $ \k sink ->
sink =<< liftIO (F.readFile =<< getLocation d k)
retrieveKeyFileCheapM :: OsPath -> ChunkConfig -> Maybe (Key -> AssociatedFile -> OsPath -> Annex ())
)
)
-storeExportM :: OsPath -> CopyCoWTried -> OsPath -> Key -> ExportLocation -> MeterUpdate -> Annex ()
-storeExportM d cow src _k loc p = do
+storeExportM :: OsPath -> CopyCoWTried -> FastCopy -> OsPath -> Key -> ExportLocation -> MeterUpdate -> Annex ()
+storeExportM d cow fastcopy src _k loc p = do
liftIO $ createDirectoryUnder [d] (takeDirectory dest)
-- Write via temp file so that checkPresentGeneric will not
-- see it until it's fully stored.
where
dest = exportPath d loc
go tmp () = void $ liftIO $
- fileCopier cow src tmp p Nothing
+ fileCopier cow fastcopy src tmp p Nothing
-retrieveExportM :: OsPath -> CopyCoWTried -> Key -> ExportLocation -> OsPath -> MeterUpdate -> Annex Verification
-retrieveExportM d cow k loc dest p =
+retrieveExportM :: OsPath -> CopyCoWTried -> FastCopy -> Key -> ExportLocation -> OsPath -> MeterUpdate -> Annex Verification
+retrieveExportM d cow fastcopy k loc dest p =
verifyKeyContentIncrementally AlwaysVerify k $ \iv ->
- void $ liftIO $ fileCopier cow src dest p iv
+ void $ liftIO $ fileCopier cow fastcopy src dest p iv
where
src = exportPath d loc
=<< R.getSymbolicLinkStatus f'
guardSameContentIdentifiers cont cids currcid
-storeExportWithContentIdentifierM :: IgnoreInodes -> OsPath -> CopyCoWTried -> OsPath -> Key -> ExportLocation -> [ContentIdentifier] -> MeterUpdate -> Annex ContentIdentifier
-storeExportWithContentIdentifierM ii dir cow src _k loc overwritablecids p = do
+storeExportWithContentIdentifierM :: IgnoreInodes -> OsPath -> CopyCoWTried -> FastCopy -> OsPath -> Key -> ExportLocation -> [ContentIdentifier] -> MeterUpdate -> Annex ContentIdentifier
+storeExportWithContentIdentifierM ii dir cow fastcopy src _k loc overwritablecids p = do
liftIO $ createDirectoryUnder [dir] destdir
withTmpFileIn destdir template $ \tmpf tmph -> do
let tmpf' = fromOsPath tmpf
liftIO $ hClose tmph
- void $ liftIO $ fileCopier cow src tmpf p Nothing
+ void $ liftIO $ fileCopier cow fastcopy src tmpf p Nothing
resetAnnexFilePerm tmpf
liftIO (R.getSymbolicLinkStatus tmpf') >>= liftIO . mkContentIdentifier ii tmpf >>= \case
Nothing -> giveup "unable to generate content identifier"
inAnnex' repo rmt st key
inAnnex' :: Git.Repo -> Remote -> State -> Key -> Annex Bool
-inAnnex' repo rmt st@(State connpool duc _ _ _) key
+inAnnex' repo rmt st@(State connpool duc _ _ _ _) key
| isP2PHttp rmt = checkp2phttp
| Git.repoIsHttp repo = checkhttp
| Git.repoIsUrl repo = checkremote
dropKey' repo r st proof key
dropKey' :: Git.Repo -> Remote -> State -> Maybe SafeDropProof -> Key -> Annex ()
-dropKey' repo r st@(State connpool duc _ _ _) proof key
+dropKey' repo r st@(State connpool duc _ _ _ _) proof key
| isP2PHttp r =
clientRemoveWithProof proof key unabletoremove r >>= \case
RemoveResultPlus True fanoutuuids ->
lockKey' repo r st key callback
lockKey' :: Git.Repo -> Remote -> State -> Key -> (VerifiedCopy -> Annex r) -> Annex r
-lockKey' repo r st@(State connpool duc _ _ _) key callback
+lockKey' repo r st@(State connpool duc _ _ _ _) key callback
| isP2PHttp r = do
showLocking r
p2pHttpClient r giveup (clientLockContent key) >>= \case
copyFromRemote'' repo r st key file dest meterupdate vc
copyFromRemote'' :: Git.Repo -> Remote -> State -> Key -> AssociatedFile -> OsPath -> MeterUpdate -> VerifyConfig -> Annex Verification
-copyFromRemote'' repo r st@(State connpool _ _ _ _) key af dest meterupdate vc
+copyFromRemote'' repo r st@(State connpool _ _ _ _ _) key af dest meterupdate vc
| isP2PHttp r = copyp2phttp
| Git.repoIsHttp repo = verifyKeyContentIncrementally vc key $ \iv -> do
gc <- Annex.getGitConfig
copyToRemote' repo r st key af o meterupdate
copyToRemote' :: Git.Repo -> Remote -> State -> Key -> AssociatedFile -> Maybe OsPath -> MeterUpdate -> Annex ()
-copyToRemote' repo r st@(State connpool duc _ _ _) key af o meterupdate
+copyToRemote' repo r st@(State connpool duc _ _ _ _) key af o meterupdate
| isP2PHttp r = prepsendwith copyp2phttp
| not $ Git.repoIsUrl repo = ifM duc
( guardUsable repo (giveup "cannot access remote") $ commitOnCleanup repo r st $
- when possible.
-}
onLocal :: State -> Annex a -> Annex a
-onLocal (State _ _ _ _ lra) = onLocal' lra
+onLocal (State _ _ _ _ _ lra) = onLocal' lra
onLocalRepo :: Git.Repo -> Annex a -> Annex a
onLocalRepo repo a = do
-- done. Also returns Verified if the key's content is verified while
-- copying it.
mkFileCopier :: Bool -> State -> Annex FileCopier
-mkFileCopier remotewanthardlink (State _ _ copycowtried _ _) = do
+mkFileCopier remotewanthardlink (State _ _ copycowtried fastcopy _ _) = do
localwanthardlink <- wantHardLink
let linker = \src dest -> R.createLink (fromOsPath src) (fromOsPath dest) >> return True
if remotewanthardlink || localwanthardlink
where
copier src dest k p check verifyconfig = do
iv <- startVerifyKeyContentIncrementally verifyconfig k
- liftIO (fileCopier copycowtried src dest p iv) >>= \case
+ liftIO (fileCopier copycowtried fastcopy src dest p iv) >>= \case
Copied -> ifM check
( finishVerifyKeyContentIncrementally iv
, do
- This returns False when the repository UUID is not as expected. -}
type DeferredUUIDCheck = Annex Bool
-data State = State Ssh.P2PShellConnectionPool DeferredUUIDCheck CopyCoWTried (Annex (Git.Repo, GitConfig)) LocalRemoteAnnex
+data State = State Ssh.P2PShellConnectionPool DeferredUUIDCheck CopyCoWTried FastCopy (Annex (Git.Repo, GitConfig)) LocalRemoteAnnex
getRepoFromState :: State -> Annex Git.Repo
-getRepoFromState (State _ _ _ a _) = fst <$> a
+getRepoFromState (State _ _ _ _ a _) = fst <$> a
#ifndef mingw32_HOST_OS
{- The config of the remote git repository, cached for speed. -}
getGitConfigFromState :: State -> Annex GitConfig
-getGitConfigFromState (State _ _ _ a _) = snd <$> a
+getGitConfigFromState (State _ _ _ _ a _) = snd <$> a
#endif
mkState :: Git.Repo -> UUID -> RemoteGitConfig -> Annex State
mkState r u gc = do
pool <- Ssh.mkP2PShellConnectionPool
copycowtried <- liftIO newCopyCoWTried
+ fastcopy <- getFastCopy gc
lra <- mkLocalRemoteAnnex r
(duc, getrepo) <- go
- return $ State pool duc copycowtried getrepo lra
+ return $ State pool duc copycowtried fastcopy getrepo lra
where
go
| remoteAnnexCheckUUID gc = return
{- git-annex configuration
-
- - Copyright 2012-2024 Joey Hess <id@joeyh.name>
+ - Copyright 2012-2025 Joey Hess <id@joeyh.name>
-
- Licensed under the GNU AGPL version 3 or higher.
-}
, annexDifferences :: Differences
, annexUsedRefSpec :: Maybe RefSpec
, annexVerify :: Bool
+ , annexFastCopy :: Bool
, annexPidLock :: Bool
, annexPidLockTimeout :: Seconds
, annexDbDir :: Maybe OsPath
, annexUsedRefSpec = either (const Nothing) Just . parseRefSpec
=<< getmaybe (annexConfig "used-refspec")
, annexVerify = getbool (annexConfig "verify") True
+ , annexFastCopy = getbool (annexConfig "fastcopy") False
, annexPidLock = getbool (annexConfig "pidlock") False
, annexPidLockTimeout = Seconds $ fromMaybe 300 $
getmayberead (annexConfig "pidlocktimeout")
, remoteAnnexPush :: Bool
, remoteAnnexReadOnly :: Bool
, remoteAnnexVerify :: Bool
+ , remoteAnnexFastCopy :: Bool
, remoteAnnexCheckUUID :: Bool
, remoteAnnexTrackingBranch :: Maybe Git.Ref
, remoteAnnexTrustLevel :: Maybe String
, remoteAnnexReadOnly = getbool ReadOnlyField False
, remoteAnnexCheckUUID = getbool CheckUUIDField True
, remoteAnnexVerify = getbool VerifyField True
+ , remoteAnnexFastCopy = getbool FastCopyField False
, remoteAnnexTrackingBranch = Git.Ref . encodeBS <$>
( notempty (getmaybe TrackingBranchField)
<|> notempty (getmaybe ExportTrackingField) -- old name
| ReadOnlyField
| CheckUUIDField
| VerifyField
+ | FastCopyField
| TrackingBranchField
| ExportTrackingField
| TrustLevelField
ReadOnlyField -> inherited True "readonly"
CheckUUIDField -> uninherited True "checkuuid"
VerifyField -> inherited True "verify"
+ FastCopyField -> inherited True "fastcopy"
TrackingBranchField -> uninherited True "tracking-branch"
ExportTrackingField -> uninherited True "export-tracking"
TrustLevelField -> uninherited True "trustlevel"
{- The cp command is used, because I hate reinventing the wheel,
- and because this allows easy access to features like cp --reflink
- - and preserving metadata. -}
+ - and preserving metadata.
+ -
+ - This uses --reflink=auto when supported, which allows for fast copies
+ - using reflinks or the copy_file_range syscall. Whatever cp thinks is
+ - best. --reflink=auto is the default of recent versions of cp, but is
+ - used explicitly to support older versions. -}
copyFileExternal :: CopyMetaData -> OsPath -> OsPath -> IO Bool
copyFileExternal meta src dest = do
-- Delete any existing dest file because an unwritable file
| otherwise = return False
where
-- Note that in coreutils 9.0, cp uses CoW by default,
- -- without needing an option. This s only needed to support
- -- older versions.
+ -- without needing an option. But, this makes it fail if it is
+ -- unable to make a CoW copy.
params = Param "--reflink=always" : copyMetaDataParams meta
{- Create a hard link if the filesystem allows it, and fall back to copying
in some edge cases, where it's likely the case than an
object was downloaded incorrectly, or when needed for security.
+* `remote.<name>.annex-fastcopy`, `annex.fastcopy`
+
+ Set to "true" to optimize copying files to/from a remote.
+
+ Setting this allows `cp` to use the `copy_file_range` system call. (When
+ the `cp` command supports it.) The benefit depends on the filesystem.
+ For example, on NFS, this can allow a server-side copy to be done. On
+ ext4, this avoids some memory transfer overhead, but the file still
+ needs to be copied. There is no benefit on btrfs, since git-annex does
+ the equivilantly fast CoW reflinking on btrfs by default.
+
+ This will do little to improve copy speed when git-annex needs to read
+ the copied file in order to verify it. So for maximum speed, as well
+ as setting this option, set `remote.<name>.annex-verify` or `annex.verify`
+ to false.
+
+ Currently, setting this prevents resuming interrupted copies where
+ they left off. An interrupted copy will be re-run from the beginning.
+
* `remote.<name>.annex-tracking-branch`
This is for use with special remotes that support exports and imports.
content of any file, even though its normal location tracking does not
indicate that it does. This will cause git-annex to try to get all file
contents from the remote. Can be useful in setting up a caching remote.
-
+
* `remote.<name>.annex-proxy`
Set to "true" to make the local repository able to act as a proxy to this
[[!meta author=ben]]
[[!tag projects/INM7]]
+
+> [[done]] --[[Joey]]
as a fallback.
However, `copy_file_range` is not necessarily inexpensive. Depending on the
-filesystem it can still need to read and write the whole file. And, rather
-than a single syscall copying the whole file, git-annex would need to call
-it repeatedly in chunks in order to display a progress bar. But, making a
-lot of syscalls against a NFS filesystem would be its own overhead.
-
-So there seems to be a tradeoff between progress display and efficiency on
-NFS. And if the goal is to maximize speed for NFS with server-side copy,
-maybe progress bars are not important enough to have in that case?
+filesystem it can still need to read and write the whole file. So when
+using it, git-annex would need to poll the size of the file in order to
+update the progress bar. Or it could call the syscall repeatedly on
+chunks of the file, but on eg NFS that would add a lot of syscalls, so
+probably more overhead.
Also, it seems likely to me that you would certainly want to turn off
annex.verify along with using `copy_file_range`, which is already a manual
--- /dev/null
+[[!comment format=mdwn
+ username="joey"
+ subject="""comment 4"""
+ date="2025-06-03T18:43:21Z"
+ content="""
+Implemented `remote.<name>.annex-fastcopy` and `annex.fastcopy`
+config settings.
+
+These do just run `cp --reflink=auto`. So when a copy was interrupted,
+it won't resume where it left off. It would be possible to improve that,
+by making git-annex use `copy_file_range` itself, starting at the point it
+was interrupted. At least for keys where isStableKey is True, since
+git-annex won't be verifying when using this. But for the
+NFS use case, if the server is doing a server-side CoW copy, it's atomic
+anyway, so no need to handle resuming.
+
+Since I don't currently know of a haskell binding to `copy_file_range` and
+since it's a low-level OS-specific thing, I have punted on handling
+resuming. This could be revisited, I just don't see the benefit in the
+current use case, and wanted to start from something relatively simple.
+"""]]
projects / git-annex.git / commitdiff