summaryrefslogtreecommitdiff
path: root/apps/buffering.c
AgeCommit message (Collapse)Author
2010-02-01Align addresses in the buffering code to STORAGE_ALIGN_MASK if the target ↵Torne Wuff
has one. The PP502x DMA controller can only deal with doing DMA to 16-byte-aligned addresses because we have inadequate control over the cache to prevent interference. Other targets may also *prefer* cacheline aligned DMAs to reduce the number of cache operations required. Almost all disk reads in buffering.c will now be suitably aligned, allowing DMA to be used on PP502x. Original change from FS#9708 by Boris Gjenero (dreamlayers). git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24440 a1c6a512-1295-4272-9138-f99709370657
2009-11-22buffering.c: cosmeticsRafaël Carré
remove tabs, remove trailing spaces, indentation git-svn-id: svn://svn.rockbox.org/rockbox/trunk@23697 a1c6a512-1295-4272-9138-f99709370657
2009-11-21Commit FS#10605 - stable playback on low memory swcodec targets by Matthias ↵Michael Giacomelli
Schneider. Should allow stable playback on targets with less then 4MB of RAM and sofware decoding such as the Sandisk Clip, c200v2, m200v4 and probably others. Fixes several problems in buffering that occured when the files to be buffered weren't much smaller then the ring buffer size. Fixes a bug where move_handle would corrupt the audio buffer when trying to copy a handle that both wrapped around the highest address in the ring buffer and overlapped part of the source and desination ranges. Moves the decision in playback.c about when to update the current track handle from audio_check_new_track to after the metadata has been buffered. Corrects several other minor pieces of code. I've logged about 100 hours without a crash on various players with this patch but its possible it breaks some combination of players and features I haven't thought to test. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@23680 a1c6a512-1295-4272-9138-f99709370657
2009-11-09Removed unneeded includesJeffrey Goode
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@23578 a1c6a512-1295-4272-9138-f99709370657
2009-10-25buffering: leave a comment to explain what is broken in the code and link to ↵Rafaël Carré
FS#10605 git-svn-id: svn://svn.rockbox.org/rockbox/trunk@23344 a1c6a512-1295-4272-9138-f99709370657
2009-10-16Rework albumart buffering internally to allow for mutliple albumart sizes.Thomas Martitz
Playback now has a few albumart slots. Anything (most importantly: skins) can obtain such a slot. The slot has fields for the size which is passed to bufopen then to image_load to buffer the albumart with the proper size. Currently there's 1 slot. We can increase it for remotes if we want. Custom statusbar will increase it. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@23209 a1c6a512-1295-4272-9138-f99709370657
2009-07-14FS#10080Nils Wallménius
* Move strncpy() from core to the pluginlib * Introduce strlcpy() and use that instead in most places (use memcpy in a few) in core and some plugins * Drop strncpy() from the codec api as no codec used it * Bump codec and plugin api versions git-svn-id: svn://svn.rockbox.org/rockbox/trunk@21863 a1c6a512-1295-4272-9138-f99709370657
2009-05-26Fix yellow when building with HAVE_ALBUMART, without HAVE_JPEG/HAVE_BMP_SCALING.Andrew Mahone
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@21092 a1c6a512-1295-4272-9138-f99709370657
2009-05-15Fix the lock contention stall during cover art load, by adding a separate ↵Andrew Mahone
modify mutex for the buffer. Operations that modify contents of a buffer entry can still proceed, but ones that add or remove buffer entries, or move them in memory, will still block. Some members of struct memory_handle also need an earlier init in bufopen to make sure that buffer stats aren't trashed. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@20932 a1c6a512-1295-4272-9138-f99709370657
2009-05-04Make JPEG and BMP scaler optional with HAVE_JPEG and HAVE_BMP_SCALING, both ↵Andrew Mahone
defined for all targets that have HAVE_ALBUMART. Disable JPEG in PictureFlow and pluginlib album art search with 32KiB plugin buffer until PictureFlow gets overlay support. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@20851 a1c6a512-1295-4272-9138-f99709370657
2009-05-01Search for, and load, JPEG album art files.Andrew Mahone
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@20837 a1c6a512-1295-4272-9138-f99709370657
2009-03-08Clean up some #includesBertrik Sikken
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@20238 a1c6a512-1295-4272-9138-f99709370657
2009-02-23Show the watermark in the "Show buffering thread" screen (in bytes).Thomas Martitz
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@20096 a1c6a512-1295-4272-9138-f99709370657
2009-02-22This should fix the occasional data aborts mainly in conjunction with ↵Antonius Hellmann
AlbumArt on PP targets. See FS#9827 for details. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@20081 a1c6a512-1295-4272-9138-f99709370657
2009-01-10Calculate watermark from bitrate and harddisk spinup time.Björn Stenberg
Use a smaller PCM buffer on targets with 2MB or less ram. (FS#9703) git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19743 a1c6a512-1295-4272-9138-f99709370657
2008-12-26Make scaler output truly pluggable, add an 8-bit greyscale output toAndrew Mahone
pluginlib for use with greylib, and add source for a test scaled bmp viewer using greylib. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19593 a1c6a512-1295-4272-9138-f99709370657
2008-12-10Use cookies for thread identification instead of pointers directly which ↵Michael Sevakis
gives a buffer against wrongly identifying a thread when the slot is recycled (which has been nagging me for awhile). A slot gets 255 uses before it repeats. Everything gets incompatible so a full update is required. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19377 a1c6a512-1295-4272-9138-f99709370657
2008-12-09resize-on-load for bitmap files on 2bpp and color targetsAndrew Mahone
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19374 a1c6a512-1295-4272-9138-f99709370657
2008-12-02Fix FS#9319 and its (probably) duplicate, FS#9576. They both describe the ↵Nicolas Pennequin
boost remaining active after a change of the dynamic playlist, an issue added in r17109. The fix is to notify the buffering thread whenever an ID3 handle is created via bufopen (as is done in the other cases of bufopen). git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19304 a1c6a512-1295-4272-9138-f99709370657
2008-11-01Apply FS#9500. This adds a storage_*() abstraction to replace ata_*(). To do ↵Frank Gevaerts
that, it also introduces sd_*, nand_*, and mmc_*. This should be a good first step to allow multi-driver targets, like the Elio (ATA/SD), or the D2 (NAND/SD). git-svn-id: svn://svn.rockbox.org/rockbox/trunk@18960 a1c6a512-1295-4272-9138-f99709370657
2008-10-16Accept FS#9480 - centralise and organise the events in the apps/ layer. Jonathan Gordon
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@18819 a1c6a512-1295-4272-9138-f99709370657
2008-10-05Fix a possible deadlock if the album art loading failsJonathan Gordon
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@18711 a1c6a512-1295-4272-9138-f99709370657
2008-07-18Add handle values to the buffering LOGFQUEUE messages for easier debugging.Nicolas Pennequin
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@18092 a1c6a512-1295-4272-9138-f99709370657
2008-07-18Fix FS#9137 and FS#8194 (the end of a track is cut off after rebuffering): ↵Nicolas Pennequin
the cause was a confusion between regular subtraction and subtraction in a ring-buffer. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@18090 a1c6a512-1295-4272-9138-f99709370657
2008-07-11Hopefully fix FS #9153, as well as another bug (in fill_buffer) that could ↵Magnus Holmgren
cause problems when rebuffering. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@18009 a1c6a512-1295-4272-9138-f99709370657
2008-07-02Fix FS#8964 (CPU remains boosted after buffering). 'filling' needs to be set ↵Nicolas Pennequin
to false when there's nothing left to buffer. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17923 a1c6a512-1295-4272-9138-f99709370657
2008-06-28Updated our source code header to explicitly mention that we are GPL v2 orDaniel Stenberg
later. We still need to hunt down snippets used that are not. 1324 modified files... http://www.rockbox.org/mail/archive/rockbox-dev-archive-2008-06/0060.shtml git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17847 a1c6a512-1295-4272-9138-f99709370657
2008-05-13Lock the linked list mutex when accessing the memory handles. This should ↵Nicolas Pennequin
fix the invalid values appearing for short times in the buffering debug screen. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17497 a1c6a512-1295-4272-9138-f99709370657
2008-04-30Remove unused #define.Steve Bavin
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17294 a1c6a512-1295-4272-9138-f99709370657
2008-04-28Cleaned up playblack.h header file and usage of it.Bertrik Sikken
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17276 a1c6a512-1295-4272-9138-f99709370657
2008-04-16Fix FS#8902 (NSF files get skipped). The cause was too strict metadata ↵Nicolas Pennequin
checking (the NSF metadata parser doesn't fill the 'length' field). Hopefully there won't be any problems with that anymore. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17142 a1c6a512-1295-4272-9138-f99709370657
2008-04-14A rather big change to how tracks are loaded: there are now two parts to the ↵Nicolas Pennequin
process and metadata loading is done by the buffering thread (except for the first unbuffered track). The audio thread now calls audio_load_track, and once the metadata is loaded, the buffering thread sends an event which will make the audio thread call audio_finish_load_track. This one then takes care of the rest of the loading. This method makes skipping noticeably faster for unbuffered tracks, and especially backwards. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17109 a1c6a512-1295-4272-9138-f99709370657
2008-04-13Make mpegplayer sleep the disk after buffering to save battery. * Add a ↵Jens Arnold
simulator stub for ata_sleep(), and un-ifdef most calls to it. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17096 a1c6a512-1295-4272-9138-f99709370657
2008-04-03Migrate the buffering code to the new events system.Nicolas Pennequin
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16950 a1c6a512-1295-4272-9138-f99709370657
2008-03-29Just change PRIORITY_BUFFERING to 15. Seems a teeny tiny bit helpful to ↵Michael Sevakis
mpegplayer as well. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16884 a1c6a512-1295-4272-9138-f99709370657
2008-03-29Change the default buffer filechunk back to 32KB and bump the buffering ↵Michael Sevakis
thread priority up by one. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16883 a1c6a512-1295-4272-9138-f99709370657
2008-03-29Fix sim warnings.Nicolas Pennequin
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16881 a1c6a512-1295-4272-9138-f99709370657
2008-03-29Boost the audio thread while it's filling the buffer. In the audio and the ↵Nicolas Pennequin
buffering threads, only cancel cpu boost when not filling. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16880 a1c6a512-1295-4272-9138-f99709370657
2008-03-29Correct some windows line endings back to unix.Nicolas Pennequin
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16877 a1c6a512-1295-4272-9138-f99709370657
2008-03-28Revert my earlier const madness, we'll keep the parameter lists simple.Steve Bavin
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16863 a1c6a512-1295-4272-9138-f99709370657
2008-03-28The const police raid playback.c, should be no change to behaviour.Steve Bavin
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16860 a1c6a512-1295-4272-9138-f99709370657
2008-03-26More consts that require no functional change.Steve Bavin
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16814 a1c6a512-1295-4272-9138-f99709370657
2008-03-25Add a complete priority inheritance implementation to the scheduler (all ↵Michael Sevakis
mutex ownership and queue_send calls are inheritable). Priorities are differential so that dispatch depends on the runnable range of priorities. Codec priority can therefore be raised in small steps (pcmbuf updated to enable). Simplify the kernel functions to ease implementation and use the same kernel.c for both sim and target (I'm tired of maintaining two ;_). 1) Not sure if a minor audio break at first buffering issue will exist on large-sector disks (the main mutex speed issue was genuinely resolved earlier). At this point it's best dealt with at the buffering level. It seems a larger filechunk could be used again. 2) Perhaps 64-bit sims will have some minor issues (finicky) but a backroll of the code of concern there is a 5-minute job. All kernel objects become incompatible so a full rebuild and update is needed. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16791 a1c6a512-1295-4272-9138-f99709370657
2008-02-12Notify the buffering thread when a handle is added, so it can go into ↵Nicolas Pennequin
filling mode. This is some sort of a replacement for the ATA idle callback. It will interrupt an ongoing buffering process (buffer_handle), but not for long enough to be a problem. Should fix a problem reported by Dave Hooper where inserting tracks into the playlist would flush the buffer and not refill it, sometimes causing the inserted tracks to be skipped. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16295 a1c6a512-1295-4272-9138-f99709370657
2008-01-08Fix FS#6215 (player crashes after trying to resume deleted music file). As ↵Nicolas Pennequin
mentioned in the lastest comment, the problem was in trying to resume at an offset larger than the size of the file following the one that was deleted. This revealed a crash in the buffering code, which gets a fix, but we also need to intervene earlier in the track loading so that the track that gets played will be from its start. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16028 a1c6a512-1295-4272-9138-f99709370657
2007-12-16Check that the pointer isn't NULL before dereferencing it, just to be safe ↵Nicolas Pennequin
and to allow using bufgetdata to know how much data is available. Also remove a few trailing spaces. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@15940 a1c6a512-1295-4272-9138-f99709370657
2007-11-28Possibly save a few bytes when stripping the tags of the last allocated handleBrandon Low
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@15844 a1c6a512-1295-4272-9138-f99709370657
2007-11-28Fix FS8069, because Nico_P made it easyBrandon Low
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@15840 a1c6a512-1295-4272-9138-f99709370657
2007-11-26Checking for total free buffer space isn't correct. What we need to check is ↵Nicolas Pennequin
the space available for the particular handle we want to rebuffer. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@15821 a1c6a512-1295-4272-9138-f99709370657
2007-11-26Buffering callbacks rework. There is now one callback for all the events ↵Nicolas Pennequin
that can occur. Callbacks are now registred only once instead of being removed after having been called. Fix FS#8092 by flushing the audio when a rebuffer is needed. Also add some comments here and there. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@15816 a1c6a512-1295-4272-9138-f99709370657
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-02 11:25:28 +0000