Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
/***************************************************************************
|
|
|
|
|
* __________ __ ___.
|
|
|
|
|
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
|
|
|
|
|
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
|
|
|
|
|
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
|
|
|
|
|
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
|
|
|
|
|
* \/ \/ \/ \/ \/
|
|
|
|
|
* $Id$
|
|
|
|
|
*
|
|
|
|
|
* Copyright (C) 2014 by Michael Sevakis
|
|
|
|
|
*
|
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
|
* modify it under the terms of the GNU General Public License
|
|
|
|
|
* as published by the Free Software Foundation; either version 2
|
|
|
|
|
* of the License, or (at your option) any later version.
|
|
|
|
|
*
|
|
|
|
|
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
|
|
|
|
|
* KIND, either express or implied.
|
|
|
|
|
*
|
|
|
|
|
****************************************************************************/
|
|
|
|
|
#include "config.h"
|
|
|
|
|
#include <errno.h>
|
|
|
|
|
#include "system.h"
|
|
|
|
|
#include "debug.h"
|
|
|
|
|
#include "panic.h"
|
|
|
|
|
#include "pathfuncs.h"
|
|
|
|
|
#include "disk_cache.h"
|
|
|
|
|
#include "fileobj_mgr.h"
|
|
|
|
|
#include "dir.h"
|
|
|
|
|
#include "dircache_redirect.h"
|
|
|
|
|
#include "dircache.h"
|
|
|
|
|
#include "string-extra.h"
|
|
|
|
|
#include "rbunicode.h"
|
|
|
|
|
|
|
|
|
|
/** Internal common filesystem service functions **/
|
|
|
|
|
|
|
|
|
|
/* for internal functions' scanning use to save quite a bit of stack space -
|
|
|
|
|
access must be serialized by the writer lock */
|
2014-08-30 03:36:11 +00:00
|
|
|
#if defined(CPU_SH) || defined(IAUDIO_M5) \
|
|
|
|
|
|| CONFIG_CPU == IMX233
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
/* otherwise, out of IRAM */
|
|
|
|
|
struct fat_direntry dir_fatent;
|
|
|
|
|
#else
|
|
|
|
|
struct fat_direntry dir_fatent IBSS_ATTR;
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
struct mrsw_lock file_internal_mrsw SHAREDBSS_ATTR;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/** File stream sector caching **/
|
|
|
|
|
|
|
|
|
|
/* initialize a new cache structure */
|
|
|
|
|
void file_cache_init(struct filestr_cache *cachep)
|
|
|
|
|
{
|
|
|
|
|
cachep->buffer = NULL;
|
|
|
|
|
cachep->sector = INVALID_SECNUM;
|
|
|
|
|
cachep->flags = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* discard and mark the cache buffer as unused */
|
|
|
|
|
void file_cache_reset(struct filestr_cache *cachep)
|
|
|
|
|
{
|
|
|
|
|
cachep->sector = INVALID_SECNUM;
|
|
|
|
|
cachep->flags = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* allocate resources attached to the cache */
|
|
|
|
|
void file_cache_alloc(struct filestr_cache *cachep)
|
|
|
|
|
{
|
|
|
|
|
/* if this fails, it is a bug; check for leaks and that the cache has
|
|
|
|
|
enough buffers for the worst case */
|
|
|
|
|
if (!cachep->buffer && !(cachep->buffer = dc_get_buffer()))
|
|
|
|
|
panicf("file_cache_alloc - OOM");
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* free resources attached to the cache */
|
|
|
|
|
void file_cache_free(struct filestr_cache *cachep)
|
|
|
|
|
{
|
|
|
|
|
if (cachep && cachep->buffer)
|
|
|
|
|
{
|
|
|
|
|
dc_release_buffer(cachep->buffer);
|
|
|
|
|
cachep->buffer = NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
file_cache_reset(cachep);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/** Stream base APIs **/
|
|
|
|
|
|
|
|
|
|
static inline void filestr_clear(struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
stream->flags = 0;
|
|
|
|
|
stream->bindp = NULL;
|
|
|
|
|
#if 0
|
|
|
|
|
stream->mtx = NULL;
|
|
|
|
|
#endif
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* actually late-allocate the assigned cache */
|
|
|
|
|
void filestr_alloc_cache(struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
file_cache_alloc(stream->cachep);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* free the stream's cache buffer if it's its own */
|
|
|
|
|
void filestr_free_cache(struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
if (stream->cachep == &stream->cache)
|
|
|
|
|
file_cache_free(stream->cachep);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* assign a cache to the stream */
|
|
|
|
|
void filestr_assign_cache(struct filestr_base *stream,
|
|
|
|
|
struct filestr_cache *cachep)
|
|
|
|
|
{
|
|
|
|
|
if (cachep)
|
|
|
|
|
{
|
|
|
|
|
filestr_free_cache(stream);
|
|
|
|
|
stream->cachep = cachep;
|
|
|
|
|
}
|
|
|
|
|
else /* assign own cache */
|
|
|
|
|
{
|
|
|
|
|
file_cache_reset(&stream->cache);
|
|
|
|
|
stream->cachep = &stream->cache;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* duplicate a cache into a stream's local cache */
|
|
|
|
|
void filestr_copy_cache(struct filestr_base *stream,
|
|
|
|
|
struct filestr_cache *cachep)
|
|
|
|
|
{
|
|
|
|
|
stream->cachep = &stream->cache;
|
|
|
|
|
stream->cache.sector = cachep->sector;
|
|
|
|
|
stream->cache.flags = cachep->flags;
|
|
|
|
|
|
|
|
|
|
if (cachep->buffer)
|
|
|
|
|
{
|
|
|
|
|
file_cache_alloc(&stream->cache);
|
|
|
|
|
memcpy(stream->cache.buffer, cachep->buffer, DC_CACHE_BUFSIZE);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
file_cache_free(&stream->cache);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* discard cache contents and invalidate it */
|
|
|
|
|
void filestr_discard_cache(struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
file_cache_reset(stream->cachep);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Initialize the base descriptor */
|
|
|
|
|
void filestr_base_init(struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
filestr_clear(stream);
|
|
|
|
|
file_cache_init(&stream->cache);
|
|
|
|
|
stream->cachep = &stream->cache;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* free base descriptor resources */
|
|
|
|
|
void filestr_base_destroy(struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
filestr_clear(stream);
|
|
|
|
|
filestr_free_cache(stream);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/** Internal directory service functions **/
|
|
|
|
|
|
|
|
|
|
/* read the next directory entry and return its FS info */
|
|
|
|
|
int uncached_readdir_internal(struct filestr_base *stream,
|
|
|
|
|
struct file_base_info *infop,
|
|
|
|
|
struct fat_direntry *fatent)
|
|
|
|
|
{
|
|
|
|
|
return fat_readdir(&stream->fatstr, &infop->fatfile.e,
|
|
|
|
|
filestr_get_cache(stream), fatent);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* rewind the FS directory to the beginning */
|
|
|
|
|
void uncached_rewinddir_internal(struct file_base_info *infop)
|
|
|
|
|
{
|
|
|
|
|
fat_rewinddir(&infop->fatfile.e);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* check if the directory is empty (ie. only "." and/or ".." entries
|
|
|
|
|
exist at most) */
|
|
|
|
|
int test_dir_empty_internal(struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
int rc;
|
|
|
|
|
|
|
|
|
|
struct file_base_info info;
|
|
|
|
|
fat_rewind(&stream->fatstr);
|
|
|
|
|
rewinddir_internal(&info);
|
|
|
|
|
|
|
|
|
|
while ((rc = readdir_internal(stream, &info, &dir_fatent)) > 0)
|
|
|
|
|
{
|
|
|
|
|
/* no OEM decoding is recessary for this simple check */
|
|
|
|
|
if (!is_dotdir_name(dir_fatent.name))
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("Directory not empty\n");
|
|
|
|
|
FILE_ERROR_RETURN(ENOTEMPTY, -1);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("I/O error checking directory: %d\n", rc);
|
|
|
|
|
FILE_ERROR_RETURN(EIO, rc * 10 - 2);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* iso decode the name to UTF-8 */
|
|
|
|
|
void iso_decode_d_name(char *d_name)
|
|
|
|
|
{
|
|
|
|
|
if (is_dotdir_name(d_name))
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
char shortname[13];
|
|
|
|
|
size_t len = strlcpy(shortname, d_name, sizeof (shortname));
|
|
|
|
|
/* This MUST be the default codepage thus not something that could be
|
|
|
|
|
loaded on call */
|
|
|
|
|
iso_decode(shortname, d_name, -1, len + 1);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#ifdef HAVE_DIRCACHE
|
|
|
|
|
/* nullify all the fields of the struct dirent */
|
|
|
|
|
void empty_dirent(struct dirent *entry)
|
|
|
|
|
{
|
|
|
|
|
entry->d_name[0] = '\0';
|
|
|
|
|
entry->info.attr = 0;
|
|
|
|
|
entry->info.size = 0;
|
|
|
|
|
entry->info.wrtdate = 0;
|
|
|
|
|
entry->info.wrttime = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* fill the native dirinfo from the static dir_fatent */
|
|
|
|
|
void fill_dirinfo_native(struct dirinfo_native *dinp)
|
|
|
|
|
{
|
|
|
|
|
struct fat_direntry *fatent = get_dir_fatent();
|
|
|
|
|
dinp->attr = fatent->attr;
|
|
|
|
|
dinp->size = fatent->filesize;
|
|
|
|
|
dinp->wrtdate = fatent->wrtdate;
|
|
|
|
|
dinp->wrttime = fatent->wrttime;
|
|
|
|
|
}
|
|
|
|
|
#endif /* HAVE_DIRCACHE */
|
|
|
|
|
|
|
|
|
|
int uncached_readdir_dirent(struct filestr_base *stream,
|
|
|
|
|
struct dirscan_info *scanp,
|
|
|
|
|
struct dirent *entry)
|
|
|
|
|
{
|
|
|
|
|
struct fat_direntry fatent;
|
|
|
|
|
int rc = fat_readdir(&stream->fatstr, &scanp->fatscan,
|
|
|
|
|
filestr_get_cache(stream), &fatent);
|
|
|
|
|
|
|
|
|
|
/* FAT driver clears the struct fat_dirent if nothing is returned */
|
|
|
|
|
strcpy(entry->d_name, fatent.name);
|
|
|
|
|
entry->info.attr = fatent.attr;
|
|
|
|
|
entry->info.size = fatent.filesize;
|
|
|
|
|
entry->info.wrtdate = fatent.wrtdate;
|
|
|
|
|
entry->info.wrttime = fatent.wrttime;
|
|
|
|
|
|
|
|
|
|
return rc;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* rewind the FS directory pointer */
|
|
|
|
|
void uncached_rewinddir_dirent(struct dirscan_info *scanp)
|
|
|
|
|
{
|
|
|
|
|
fat_rewinddir(&scanp->fatscan);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/** open_stream_internal() helpers and types **/
|
|
|
|
|
|
|
|
|
|
struct pathwalk
|
|
|
|
|
{
|
|
|
|
|
const char *path; /* current location in input path */
|
|
|
|
|
unsigned int callflags; /* callflags parameter */
|
|
|
|
|
struct path_component_info *compinfo; /* compinfo parameter */
|
|
|
|
|
file_size_t filesize; /* size of the file */
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
struct pathwalk_component
|
|
|
|
|
{
|
|
|
|
|
struct file_base_info info; /* basic file information */
|
|
|
|
|
const char *name; /* component name location in path */
|
|
|
|
|
uint16_t length; /* length of name of component */
|
|
|
|
|
uint16_t attr; /* attributes of this component */
|
|
|
|
|
struct pathwalk_component *nextp; /* parent if in use else next free */
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
#define WALK_RC_NOT_FOUND 0 /* successfully not found */
|
|
|
|
|
#define WALK_RC_FOUND 1 /* found and opened */
|
|
|
|
|
#define WALK_RC_FOUND_ROOT 2 /* found and opened sys/volume root */
|
|
|
|
|
#define WALK_RC_CONT_AT_ROOT 3 /* continue at root level */
|
|
|
|
|
|
|
|
|
|
/* return another struct pathwalk_component from the pool, or NULL if the
|
|
|
|
|
pool is completely used */
|
|
|
|
|
static void * pathwalk_comp_alloc_(struct pathwalk_component *parentp)
|
|
|
|
|
{
|
|
|
|
|
/* static pool that goes to a depth of STATIC_COMP_NUM before allocating
|
|
|
|
|
elements from the stack */
|
|
|
|
|
static struct pathwalk_component aux_pathwalk[STATIC_PATHCOMP_NUM];
|
|
|
|
|
struct pathwalk_component *compp = NULL;
|
|
|
|
|
|
|
|
|
|
if (!parentp)
|
|
|
|
|
compp = &aux_pathwalk[0]; /* root */
|
|
|
|
|
else if (PTR_IN_ARRAY(aux_pathwalk, parentp, STATIC_PATHCOMP_NUM-1))
|
|
|
|
|
compp = parentp + 1;
|
|
|
|
|
|
|
|
|
|
return compp;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* allocates components from the pool or stack depending upon the depth */
|
|
|
|
|
#define pathwalk_comp_alloc(parentp) \
|
|
|
|
|
({ \
|
|
|
|
|
void *__c = pathwalk_comp_alloc_(parentp); \
|
|
|
|
|
if (!__builtin_constant_p(parentp) && !__c) \
|
|
|
|
|
__c = alloca(sizeof (struct pathwalk_component)); \
|
|
|
|
|
(struct pathwalk_component *)__c; \
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
/* fill in the details of the struct path_component_info for caller */
|
|
|
|
|
static int fill_path_compinfo(struct pathwalk *walkp,
|
|
|
|
|
struct pathwalk_component *compp,
|
|
|
|
|
int rc)
|
|
|
|
|
{
|
|
|
|
|
if (rc == -ENOENT)
|
|
|
|
|
{
|
|
|
|
|
/* this component wasn't found; see if more of them exist or path
|
|
|
|
|
has trailing separators; if it does, this component should be
|
|
|
|
|
interpreted as a directory even if it doesn't exist and it's the
|
|
|
|
|
final one; also, this has to be the last part or it's an error*/
|
|
|
|
|
const char *p = GOBBLE_PATH_SEPCH(walkp->path);
|
|
|
|
|
if (!*p)
|
|
|
|
|
{
|
|
|
|
|
if (p > walkp->path)
|
|
|
|
|
compp->attr |= ATTR_DIRECTORY;
|
|
|
|
|
|
|
|
|
|
rc = WALK_RC_NOT_FOUND; /* successfully not found */
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (rc >= 0)
|
|
|
|
|
{
|
|
|
|
|
struct path_component_info *compinfo = walkp->compinfo;
|
|
|
|
|
compinfo->name = compp->name;
|
|
|
|
|
compinfo->length = compp->length;
|
|
|
|
|
compinfo->attr = compp->attr;
|
|
|
|
|
compinfo->filesize = walkp->filesize;
|
|
|
|
|
compinfo->parentinfo = (compp->nextp ?: compp)->info;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return rc;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* open the final stream itself, if found */
|
|
|
|
|
static int walk_open_info(struct pathwalk *walkp,
|
|
|
|
|
struct pathwalk_component *compp,
|
|
|
|
|
int rc,
|
|
|
|
|
struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
/* this may make adjustments to things; do it first */
|
|
|
|
|
if (walkp->compinfo)
|
|
|
|
|
rc = fill_path_compinfo(walkp, compp, rc);
|
|
|
|
|
|
|
|
|
|
if (rc < 0 || rc == WALK_RC_NOT_FOUND)
|
|
|
|
|
return rc;
|
|
|
|
|
|
|
|
|
|
unsigned int callflags = walkp->callflags;
|
|
|
|
|
bool isdir = compp->attr & ATTR_DIRECTORY;
|
|
|
|
|
|
|
|
|
|
/* type must match what is called for */
|
|
|
|
|
switch (callflags & FF_TYPEMASK)
|
|
|
|
|
{
|
|
|
|
|
case FF_FILE:
|
|
|
|
|
if (!isdir) break;
|
|
|
|
|
DEBUGF("File is a directory\n");
|
|
|
|
|
return -EISDIR;
|
|
|
|
|
case FF_DIR:
|
|
|
|
|
if (isdir) break;
|
|
|
|
|
DEBUGF("File is not a directory\n");
|
|
|
|
|
return -ENOTDIR;
|
|
|
|
|
/* FF_ANYTYPE: basically, ignore FF_FILE/FF_DIR */
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* FO_DIRECTORY must match type */
|
|
|
|
|
if (isdir)
|
|
|
|
|
callflags |= FO_DIRECTORY;
|
|
|
|
|
else
|
|
|
|
|
callflags &= ~FO_DIRECTORY;
|
|
|
|
|
|
|
|
|
|
fileop_onopen_internal(stream, &compp->info, callflags);
|
|
|
|
|
return compp->nextp ? WALK_RC_FOUND : WALK_RC_FOUND_ROOT;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* check the component against the prefix test info */
|
|
|
|
|
static void walk_check_prefix(struct pathwalk *walkp,
|
|
|
|
|
struct pathwalk_component *compp)
|
|
|
|
|
{
|
|
|
|
|
if (compp->attr & ATTR_PREFIX)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
if (!fat_file_is_same(&compp->info.fatfile,
|
|
|
|
|
&walkp->compinfo->prefixp->fatfile))
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
compp->attr |= ATTR_PREFIX;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* opens the component named by 'comp' in the directory 'parent' */
|
|
|
|
|
static NO_INLINE int open_path_component(struct pathwalk *walkp,
|
|
|
|
|
struct pathwalk_component *compp,
|
|
|
|
|
struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
int rc;
|
|
|
|
|
|
|
|
|
|
/* create a null-terminated copy of the component name */
|
|
|
|
|
char *compname = strmemdupa(compp->name, compp->length);
|
|
|
|
|
|
|
|
|
|
unsigned int callflags = walkp->callflags;
|
|
|
|
|
struct pathwalk_component *parentp = compp->nextp;
|
|
|
|
|
|
|
|
|
|
/* children inherit the prefix coloring from the parent */
|
|
|
|
|
compp->attr = parentp->attr & ATTR_PREFIX;
|
|
|
|
|
|
|
|
|
|
/* most of the next would be abstracted elsewhere if doing other
|
|
|
|
|
filesystems */
|
|
|
|
|
|
|
|
|
|
/* scan parent for name; stream is converted to this parent */
|
|
|
|
|
file_cache_reset(stream->cachep);
|
|
|
|
|
stream->infop = &parentp->info;
|
|
|
|
|
fat_filestr_init(&stream->fatstr, &parentp->info.fatfile);
|
|
|
|
|
rewinddir_internal(&compp->info);
|
|
|
|
|
|
|
|
|
|
while ((rc = readdir_internal(stream, &compp->info, &dir_fatent)) > 0)
|
|
|
|
|
{
|
|
|
|
|
if (rc > 1 && !(callflags & FF_NOISO))
|
|
|
|
|
iso_decode_d_name(dir_fatent.name);
|
|
|
|
|
|
|
|
|
|
if (!strcasecmp(compname, dir_fatent.name))
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (rc == 0)
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("File/directory not found\n");
|
|
|
|
|
return -ENOENT;
|
|
|
|
|
}
|
|
|
|
|
else if (rc < 0)
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("I/O error reading directory %d\n", rc);
|
|
|
|
|
return -EIO;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
rc = fat_open(stream->fatstr.fatfilep, dir_fatent.firstcluster,
|
|
|
|
|
&compp->info.fatfile);
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("I/O error opening file/directory %s (%d)\n",
|
|
|
|
|
compname, rc);
|
|
|
|
|
return -EIO;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
walkp->filesize = dir_fatent.filesize;
|
|
|
|
|
compp->attr |= dir_fatent.attr;
|
|
|
|
|
|
|
|
|
|
if (callflags & FF_CHECKPREFIX)
|
|
|
|
|
walk_check_prefix(walkp, compp);
|
|
|
|
|
|
|
|
|
|
return WALK_RC_FOUND;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* parse a path component, open it and process the next */
|
|
|
|
|
static NO_INLINE int
|
|
|
|
|
walk_path(struct pathwalk *walkp, struct pathwalk_component *compp,
|
|
|
|
|
struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
int rc = WALK_RC_FOUND;
|
|
|
|
|
|
|
|
|
|
if (walkp->callflags & FF_CHECKPREFIX)
|
|
|
|
|
walk_check_prefix(walkp, compp);
|
|
|
|
|
|
|
|
|
|
/* alloca is used in a loop, but we reuse any blocks previously allocated
|
|
|
|
|
if we went up then back down; if the path takes us back to the root, then
|
|
|
|
|
everything is cleaned automatically */
|
|
|
|
|
struct pathwalk_component *freep = NULL;
|
|
|
|
|
|
|
|
|
|
const char *name;
|
|
|
|
|
ssize_t len;
|
|
|
|
|
|
|
|
|
|
while ((len = parse_path_component(&walkp->path, &name)))
|
|
|
|
|
{
|
|
|
|
|
/* whatever is to be a parent must be a directory */
|
|
|
|
|
if (!(compp->attr & ATTR_DIRECTORY))
|
|
|
|
|
return -ENOTDIR;
|
|
|
|
|
|
2014-09-02 19:57:18 +00:00
|
|
|
if (len >= MAX_NAME)
|
|
|
|
|
return -ENAMETOOLONG;
|
|
|
|
|
|
|
|
|
|
/* check for "." and ".." */
|
|
|
|
|
if (name[0] == '.')
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
{
|
2014-09-02 19:57:18 +00:00
|
|
|
if (len == 1)
|
|
|
|
|
continue; /* is "." */
|
|
|
|
|
|
|
|
|
|
if (len == 2 && name[1] == '.')
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
{
|
2014-09-02 19:57:18 +00:00
|
|
|
/* is ".." */
|
|
|
|
|
struct pathwalk_component *parentp = compp->nextp;
|
|
|
|
|
if (!parentp)
|
|
|
|
|
return WALK_RC_CONT_AT_ROOT;
|
|
|
|
|
|
|
|
|
|
compp->nextp = freep;
|
|
|
|
|
freep = compp;
|
|
|
|
|
compp = parentp;
|
|
|
|
|
continue;
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
}
|
2014-09-02 19:57:18 +00:00
|
|
|
}
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
|
2014-09-02 19:57:18 +00:00
|
|
|
struct pathwalk_component *newp = freep;
|
|
|
|
|
if (!newp)
|
|
|
|
|
newp = pathwalk_comp_alloc(compp);
|
|
|
|
|
else
|
|
|
|
|
freep = freep->nextp;
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
|
2014-09-02 19:57:18 +00:00
|
|
|
newp->nextp = compp;
|
|
|
|
|
compp = newp;
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
|
2014-09-02 19:57:18 +00:00
|
|
|
compp->name = name;
|
|
|
|
|
compp->length = len;
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
|
2014-09-02 19:57:18 +00:00
|
|
|
rc = open_path_component(walkp, compp, stream);
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
break; /* return info below */
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return walk_open_info(walkp, compp, rc, stream);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* open a stream given a path to the resource */
|
|
|
|
|
int open_stream_internal(const char *path, unsigned int callflags,
|
|
|
|
|
struct filestr_base *stream,
|
|
|
|
|
struct path_component_info *compinfo)
|
|
|
|
|
{
|
2015-10-09 18:45:01 +00:00
|
|
|
DEBUGF("%s(path=\"%s\",flg=%X,str=%p,compinfo=%p)\n", __func__,
|
|
|
|
|
path, callflags, stream, compinfo);
|
Rewrite filesystem code (WIP)
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
2013-08-06 02:02:45 +00:00
|
|
|
int rc;
|
|
|
|
|
|
|
|
|
|
filestr_base_init(stream);
|
|
|
|
|
|
|
|
|
|
if (!path_is_absolute(path))
|
|
|
|
|
{
|
|
|
|
|
/* while this supports relative components, there is currently no
|
|
|
|
|
current working directory concept at this level by which to
|
|
|
|
|
fully qualify the path (though that would not be excessively
|
|
|
|
|
difficult to add) */
|
|
|
|
|
DEBUGF("\"%s\" is not an absolute path\n"
|
|
|
|
|
"Only absolute paths currently supported.\n", path);
|
|
|
|
|
FILE_ERROR(path ? ENOENT : EFAULT, -1);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* if !compinfo, then the result of this check is not visible anyway */
|
|
|
|
|
if (!compinfo)
|
|
|
|
|
callflags &= ~FF_CHECKPREFIX;
|
|
|
|
|
|
|
|
|
|
struct pathwalk walk;
|
|
|
|
|
walk.path = path;
|
|
|
|
|
walk.callflags = callflags;
|
|
|
|
|
walk.compinfo = compinfo;
|
|
|
|
|
walk.filesize = 0;
|
|
|
|
|
|
|
|
|
|
struct pathwalk_component *rootp = pathwalk_comp_alloc(NULL);
|
|
|
|
|
rootp->nextp = NULL;
|
|
|
|
|
rootp->attr = ATTR_DIRECTORY;
|
|
|
|
|
|
|
|
|
|
#ifdef HAVE_MULTIVOLUME
|
|
|
|
|
int volume = 0, rootrc = WALK_RC_FOUND;
|
|
|
|
|
#endif /* HAVE_MULTIVOLUME */
|
|
|
|
|
|
|
|
|
|
while (1)
|
|
|
|
|
{
|
|
|
|
|
const char *pathptr = walk.path;
|
|
|
|
|
|
|
|
|
|
#ifdef HAVE_MULTIVOLUME
|
|
|
|
|
/* this seamlessly integrates secondary filesystems into the
|
|
|
|
|
root namespace (e.g. "/<0>/../../<1>/../foo/." :<=> "/foo") */
|
|
|
|
|
const char *p;
|
|
|
|
|
volume = path_strip_volume(pathptr, &p, false);
|
|
|
|
|
if (!CHECK_VOL(volume))
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("No such device or address: %d\n", volume);
|
|
|
|
|
FILE_ERROR(ENXIO, -2);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* the root of this subpath is the system root? */
|
|
|
|
|
rootrc = p == pathptr ? WALK_RC_FOUND_ROOT : WALK_RC_FOUND;
|
|
|
|
|
walk.path = p;
|
|
|
|
|
#endif /* HAVE_MULTIVOLUME */
|
|
|
|
|
|
|
|
|
|
/* set name to start at last leading separator; names of volume
|
|
|
|
|
specifiers will be returned as "/<fooN>" */
|
|
|
|
|
rootp->name = GOBBLE_PATH_SEPCH(pathptr) - 1;
|
|
|
|
|
rootp->length =
|
|
|
|
|
IF_MV( rootrc == WALK_RC_FOUND ? p - rootp->name : ) 1;
|
|
|
|
|
|
|
|
|
|
rc = fat_open_rootdir(IF_MV(volume,) &rootp->info.fatfile);
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
{
|
|
|
|
|
/* not mounted */
|
|
|
|
|
DEBUGF("No such device or address: %d\n", IF_MV_VOL(volume));
|
|
|
|
|
rc = -ENXIO;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
get_rootinfo_internal(&rootp->info);
|
|
|
|
|
rc = walk_path(&walk, rootp, stream);
|
|
|
|
|
if (rc != WALK_RC_CONT_AT_ROOT)
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
switch (rc)
|
|
|
|
|
{
|
|
|
|
|
case WALK_RC_FOUND_ROOT:
|
|
|
|
|
IF_MV( rc = rootrc; )
|
|
|
|
|
case WALK_RC_NOT_FOUND:
|
|
|
|
|
case WALK_RC_FOUND:
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
default: /* utter, abject failure :`( */
|
|
|
|
|
DEBUGF("Open failed: rc=%d, errno=%d\n", rc, errno);
|
|
|
|
|
filestr_base_destroy(stream);
|
|
|
|
|
FILE_ERROR(-rc, -2);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
file_cache_reset(stream->cachep);
|
|
|
|
|
|
|
|
|
|
file_error:
|
|
|
|
|
return rc;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* close the stream referenced by 'stream' */
|
|
|
|
|
int close_stream_internal(struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
int rc;
|
|
|
|
|
unsigned int foflags = fileobj_get_flags(stream);
|
|
|
|
|
|
|
|
|
|
if ((foflags & (FO_SINGLE|FO_REMOVED)) == (FO_SINGLE|FO_REMOVED))
|
|
|
|
|
{
|
|
|
|
|
/* nothing is referencing it so now remove the file's data */
|
|
|
|
|
rc = fat_remove(&stream->infop->fatfile, FAT_RM_DATA);
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("I/O error removing file data: %d\n", rc);
|
|
|
|
|
FILE_ERROR(EIO, rc * 10 - 1);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
rc = 0;
|
|
|
|
|
file_error:
|
|
|
|
|
/* close no matter what */
|
|
|
|
|
fileop_onclose_internal(stream);
|
|
|
|
|
return rc;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* create a new stream in the parent directory */
|
|
|
|
|
int create_stream_internal(struct file_base_info *parentinfop,
|
|
|
|
|
const char *basename, size_t length,
|
|
|
|
|
unsigned int attr, unsigned int callflags,
|
|
|
|
|
struct filestr_base *stream)
|
|
|
|
|
{
|
|
|
|
|
/* assumes an attempt was made beforehand to open *stream with
|
|
|
|
|
open_stream_internal() which returned zero (successfully not found),
|
|
|
|
|
so does not initialize it here */
|
|
|
|
|
const char * const name = strmemdupa(basename, length);
|
|
|
|
|
DEBUGF("Creating \"%s\"\n", name);
|
|
|
|
|
|
|
|
|
|
struct file_base_info info;
|
|
|
|
|
int rc = fat_create_file(&parentinfop->fatfile, name, attr,
|
|
|
|
|
&info.fatfile, get_dir_fatent_dircache());
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("Create failed: %d\n", rc);
|
|
|
|
|
FILE_ERROR(rc == FAT_RC_ENOSPC ? ENOSPC : EIO, rc * 10 - 1);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* dir_fatent is implicit arg */
|
|
|
|
|
fileop_oncreate_internal(stream, &info, callflags, parentinfop, name);
|
|
|
|
|
rc = 0;
|
|
|
|
|
file_error:
|
|
|
|
|
return rc;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* removes files and directories - back-end to remove() and rmdir() */
|
|
|
|
|
int remove_stream_internal(const char *path, struct filestr_base *stream,
|
|
|
|
|
unsigned int callflags)
|
|
|
|
|
{
|
|
|
|
|
/* Only FF_* flags should be in callflags */
|
|
|
|
|
int rc;
|
|
|
|
|
|
|
|
|
|
struct filestr_base opened_stream;
|
|
|
|
|
if (!stream)
|
|
|
|
|
stream = &opened_stream;
|
|
|
|
|
|
|
|
|
|
if (stream == &opened_stream)
|
|
|
|
|
{
|
|
|
|
|
/* no stream provided so open local one */
|
|
|
|
|
rc = open_stream_internal(path, callflags, stream, NULL);
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("Failed opening path: %d\n", rc);
|
|
|
|
|
FILE_ERROR(ERRNO, rc * 10 - 1);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
/* else ignore the 'path' argument */
|
|
|
|
|
|
|
|
|
|
if (callflags & FF_DIR)
|
|
|
|
|
{
|
|
|
|
|
/* directory to be removed must be empty */
|
|
|
|
|
rc = test_dir_empty_internal(stream);
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
FILE_ERROR(ERRNO, rc * 10 - 2);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* save old info since fat_remove() will destroy the dir info */
|
|
|
|
|
struct file_base_info oldinfo = *stream->infop;
|
|
|
|
|
rc = fat_remove(&stream->infop->fatfile, FAT_RM_DIRENTRIES);
|
|
|
|
|
if (rc < 0)
|
|
|
|
|
{
|
|
|
|
|
DEBUGF("I/O error removing dir entries: %d\n", rc);
|
|
|
|
|
FILE_ERROR(EIO, rc * 10 - 3);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fileop_onremove_internal(stream, &oldinfo);
|
|
|
|
|
|
|
|
|
|
rc = 0;
|
|
|
|
|
file_error:
|
|
|
|
|
if (stream == &opened_stream)
|
|
|
|
|
{
|
|
|
|
|
/* will do removal of data below if this is the only reference */
|
|
|
|
|
int rc2 = close_stream_internal(stream);
|
|
|
|
|
if (rc2 < 0 && rc >= 0)
|
|
|
|
|
{
|
|
|
|
|
rc = rc2 * 10 - 4;
|
|
|
|
|
DEBUGF("Success but failed closing stream: %d\n", rc);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return rc;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* test file/directory existence with constraints */
|
|
|
|
|
int test_stream_exists_internal(const char *path, unsigned int callflags)
|
|
|
|
|
{
|
|
|
|
|
/* only FF_* flags should be in callflags */
|
|
|
|
|
struct filestr_base stream;
|
|
|
|
|
int rc = open_stream_internal(path, callflags, &stream, NULL);
|
|
|
|
|
if (rc > 0)
|
|
|
|
|
close_stream_internal(&stream);
|
|
|
|
|
|
|
|
|
|
return rc;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* one-time init at startup */
|
|
|
|
|
void filesystem_init(void)
|
|
|
|
|
{
|
|
|
|
|
mrsw_init(&file_internal_mrsw);
|
|
|
|
|
dc_init();
|
|
|
|
|
fileobj_mgr_init();
|
|
|
|
|
}
|