Age | Commit message (Collapse) | Author |
|
This is to avoid lingering files on windows.
|
|
The file module communicates with a file io server with the following
protocol for file operations:
> {file_request,From,ReplyAs,Request}
< {file_reply,ReplyAs,Reply}
The ReplyAs value is sent by the client side to match against when
receiving the reply and is otherwise left untouched and passed as is by
the server.
This commit enables receive optimizations by using the reference of the
server monitor, changing the protocol to:
> {file_request,From,MonitorRef,Request}
< {file_reply,MonitorRef,Reply}
As the shape of the messages is not changed, backwards compatibility is
not a concern.
|
|
* hb/file_name_type/OTP-10852:
Introduce new type file:name_all()
|
|
|
|
When the run-time system was started with +fnue, the error tuple
indicating a non-translatable filename was added as a non-proper
list tail inside an {ok,Files} term.
|
|
6d516de001dde82c02fe050db8e3aab47914fa90 added prim_file:list_dir_all/1.
Unfortunately, only the first element in the list would be handled
as intended.
|
|
|
|
|
|
When using the async thread pool and compressed files, when
an efile driver port instance is shutdown, the efile_drv
stop callback closes the file descriptor (a gzFile instance
actually) - this is dangerous if at the same time there's
an async thread performing an operation against the file,
for example calling invoke_read(), which can result in a
segmentation fault, or calling invoke_close() which double
closes the gzFile and this in turn causes 2 consecutive calls
to driver_free() against same gzFile instance (resulting in
later unexpected crashes in erl_bestfit_alloc.c for example).
The following test program makes the emulator crash when using
the async thread pool:
-module(t2).
-export([t/1]).
t(N) ->
file:delete("foo.bar"),
% Use of 'compressed' option, for creating/writing the file,
% is irrelevant. It only matters when opening it later for
% reads - a non-compressed file open with the 'compressed'
% option goes through an internal gzFile handle (instead of
% a plain integer fd), just like a compressed file.
%{ok, Fd} = file:open("foo.bar", [raw, write, binary]),
{ok, Fd} = file:open("foo.bar", [raw, write, binary, compressed]),
ok = file:write(Fd, <<"qwerty">>),
ok = file:close(Fd),
Pid = spawn_link(fun() ->
process_flag(trap_exit, true),
loop(N)
end),
Ref = erlang:monitor(process, Pid),
receive
{'DOWN', Ref, _, _, _} ->
ok
end.
loop(0) ->
ok;
loop(N) ->
Server = self(),
Pid = spawn(fun() ->
{ok, Fd} = file:open("foo.bar", [read, raw, binary, compressed]),
Server ! continue,
% Comment the file:read/2 call to make the file:close/1 call much
% more likely to crash or end up causing efile_drv to close twice
% the fd (gzFile), which will make the emulator crash later in the
% best fit allocator (erl_bestfit_alloc.c).
_ = file:read(Fd, 5),
file:close(Fd)
end),
receive continue -> ok end,
exit(Pid, shutdown),
loop(N - 1).
Running this test when using the async thread pool:
shell> erl +A 4
Erlang R15B03 (erts-5.9.3.1) [source] [64-bit] [smp:4:4] [async-threads:4] [hipe] [kernel-poll:false]
Eshell V5.9.3.1 (abort with ^G)
1> c(t2).
{ok,t2}
2> t2:t(500000).
Segmentation fault (core dumped)
When not using the async thread pool, there are no issues:
shell> erl
Erlang R15B03 (erts-5.9.3.1) [source] [64-bit] [smp:4:4] [async-threads:0] [hipe] [kernel-poll:false]
Eshell V5.9.3.1 (abort with ^G)
1> c(t2).
{ok,t2}
2> t2:t(500000).
ok
3>
An example stack trace when the crash happens because there's
an ongoing read operation is:
Thread 1 (Thread 0x7f021cf2c700 (LWP 10687)):
#0 updatewindow (strm=0x2691bf8, out=5) at zlib/inflate.c:338
#1 0x00000000005a2ba0 in inflate (strm=0x2691bf8, flush=0) at zlib/inflate.c:1141
#2 0x000000000055c46a in erts_gzread (file=0x2691bf8, buf=0x7f0215b29e80, len=5) at drivers/common/gzio.c:523
#3 0x00000000005849ef in invoke_read (data=0x26b2228) at drivers/common/efile_drv.c:1114
#4 0x000000000050adcb in async_main (arg=0x7f021bf5cf40) at beam/erl_async.c:488
#5 0x00000000005c21a0 in thr_wrapper (vtwd=0x7fff69c6ff10) at pthread/ethread.c:106
#6 0x00007f021c573e9a in start_thread () from /lib/x86_64-linux-gnu/libpthread.so.0
#7 0x00007f021c097cbd in clone () from /lib/x86_64-linux-gnu/libc.so.6
#8 0x0000000000000000 in ?? ()
And when there's an ongoing close operation when the driver
is stopped:
Thread 1 (Thread 0x7fe5f5654700 (LWP 747)):
#0 0x0000000000459b64 in bf_unlink_free_block (block=0x10b2a70, allctr=<optimized out>, flags=<optimized out>) at beam/erl_bestfit_alloc.c:792
#1 bf_unlink_free_block (flags=0, block=0x10b2a70, allctr=0x873380) at beam/erl_bestfit_alloc.c:822
#2 bf_get_free_block (allctr=0x873380, size=<optimized out>, cand_blk=<optimized out>, cand_size=<optimized out>, flags=0) at beam/erl_bestfit_alloc.c:869
#3 0x000000000044f0dd in mbc_alloc_block (alcu_flgsp=<synthetic pointer>, blk_szp=<synthetic pointer>, size=200, allctr=0x873380) at beam/erl_alloc_util.c:1198
#4 mbc_alloc (allctr=0x873380, size=200) at beam/erl_alloc_util.c:1345
#5 0x000000000045449b in do_erts_alcu_alloc (size=200, extra=0x873380, type=165) at beam/erl_alloc_util.c:3442
#6 erts_alcu_alloc_thr_pref (type=165, extra=<optimized out>, size=192) at beam/erl_alloc_util.c:3520
#7 0x000000000055c0bf in gz_open (mode=0x5d98b2 "rb", path=0x1103418 "foo.bar") at drivers/common/gzio.c:164
#8 erts_gzopen (path=0x1103418 "foo.bar", mode=0x5d98b2 "rb") at drivers/common/gzio.c:307
#9 0x0000000000584e47 in invoke_open (data=0x1103330) at drivers/common/efile_drv.c:1857
#10 0x000000000050adcb in async_main (arg=0x7fe5f698af80) at beam/erl_async.c:488
|
|
This operation allows pre-allocation of space for files.
It succeeds only on systems that support such operation.
The POSIX standard defines the optional system call
posix_fallocate() to implement this feature. However,
some systems implement more specific functions to
accomplish the same operation.
On Linux, if the more specific function fallocate() is
implemented, it is used instead of posix_fallocate(),
falling back to posix_fallocate() if the fallocate()
call failed (it's only supported for the ext4, ocfs2,
xfs and btrfs file systems at the moment).
On Mac OS X it uses the specific fcntl() operation
F_PREALLOCATE, falling back to posix_fallocate() if
it's available (at the moment Mac OS X doesn't provide
posix_fallocate()).
On any other UNIX system, it uses posix_fallocate() if it's
available. Any other system not providing this system call
or any function to pre-allocate space for files, this operation
always fails with the ENOTSUP POSIX error.
|
|
|
|
Long input paths (longer than MAX_PATH) would get copied
into a buffer of size MAX_PATH for read_link and altname
in efile_drv.
Also fixed misuse of size_t parameter as wchar_t *
string length in win_efile:efile_readlink.
|
|
|
|
|
|
Temporary workaround until there is a memory info function
conveniantly usable from test suites.
|
|
Windows needs large timeout.
|
|
|
|
|
|
Test writing 2GB to a file in one go.
|
|
Factor out support functions from large_file/1 so that we can
write other test cases that use large files.
|
|
unix_free/1 returned the total size of the disk, not the free
space.
|
|
|
|
|
|
|
|
|
|
The gzio driver goes into an infinite loop when reading past the end of
a compressed file.
Reported-By: Alex Morarash
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Add an option that atomically tests for the existence of a file and
creates it if the file does not exist, by passing the O_EXCL flag
to open() on Unix and CREATE_NEW flag on Windows. Support for O_EXCL
varies across platforms and filesystems.
{ok, Fd} = file:open("/tmp/foo", [write,exclusive]),
{error, eexist} = file:open("/tmp/foo", [write,exclusive]).
|
|
|
|
Useful for informing the Operating System about the access pattern
for a file's data, so that it can adapt the caching strategy to
maximize disk IO performance.
|
|
file:datasync/1 invokes the POSIX system call "int fdatasync(int fd)".
This system call is similar to "fsync" but, unlike fsync, it does not
update the metadata associated with the file (like the access time for
example). It's used by many DBMSs (MySQL and SQLite of example) to
increase disk IO performance, as it avoids disk seeks and disk write
operations compared to fsync.
More details on it at:
http://linux.die.net/man/2/fdatasync
An example, from the MySQL source:
http://bazaar.launchpad.net/~mysql/mysql-server/mysql-5.1-telco-6.1/annotate/head%3A/mysys/my_sync.c#L61
This new function just calls fsync on systems not implementing fdatasync.
|
|
On FreeBSD, file:del_dir("..") will return {error,einval} rather
than the expected {error,eexist}, and so will file:del_dir("../.."),
and so on.
It could be argued that we should change the implementation of
file:del_dir/1 to remap the error code (as some other error codes
are remapped to reduce the differences between different platforms),
but the consistency gained does not seem to be worth the effort.
Therefore, until we'll find a real-world use case where it is
essential to have consistent error codes for file:del_dir("..") on
all platforms, change the test case to accept both errors.
|
|
* bg/cleanup-tests:
file_SUITE: eliminate a warning for an unused variable
kernel tests: modernize guard tests
unicode_SUITE: replace deprecated concat_binary/1 with list_to_binary/1
stdlib tests: modernize guard tests
Test suites: fix creation of Emakefiles
|
|
|
|
|