compress: Block if unable to write all data in single write()
Previously compress_write() would only block on poll() if the
available space < fragment_size. If the device has a small fragment
size this could lead to it never blocking and instead looping around
doing many small writes. This is bad for power saving.
If we were unable to write all the remaining data in a single write
we want to block until the device reaches a buffer high water mark,
to allow the CPU to sleep.
This change will always attempt to issue the first write providing
avail >= fragment_size. All subsequent loops will block on poll()
before attempting another write() unless there is enough buffer space
to write all remaining data.
compress: Allow client to pass "don't care" for fragment size/number
Client cannot know what is a sensible fragment size and count
for the device unless it has prior knowledge of the precise
hardware. To allow for generic clients not hardcoded for particular
hardware, this change allows fragment_size and fragments to be
passed as 0, meaning "don't care" and tinycompress will choose
default values read from the driver.
compress: Handle case of pause() during compress_write()
If stream is paused() during a compress_write() the poll()
will return 0 or -EBADFD, or the write() will return -EBADFD.
This is not an error so compress_write() should not pass an
error code to the caller. It should abort the write and then
return the number of bytes it had written up to that point.
compress: Remove hardcoded limit on length of poll() wait
For best power-saving we want to sleep on poll() for as long as
possible, we don't want to wake unnecessarily for arbitrary time
limits. Adds function compress_set_max_poll_wait() so that
client can configure the poll() timeout.
compress: Change running from bitfield to plain int
Declaring running as a single-bit bitfield does not save any space
or give any coding advantages, but does leave potential opportunities
for the compiler to be inefficient in how it handles bitfields compared
to how it would handle a machine int.
compress: Add function to get timestamp in samples
The compress_get_hpointer() converts the timestamp into actual
time. But Android needs it in samples. To avoid the inefficiency
of using compress_get_hpointer() and converting into time and then
back into samples, this change adds compress_get_tstamp() which
returns the raw sample count.
Charles Keepax [Fri, 25 Jan 2013 10:43:14 +0000 (10:43 +0000)]
compress: Do not put newlines on error messages
In the oops function text is appended onto the end of the supplied error
message, so newlines appear in the middle of error messages. This patch
removes all newlines from supplied error messages.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Charles Keepax [Fri, 25 Jan 2013 10:43:13 +0000 (10:43 +0000)]
compress: Return error messages correctly from compress_open
The allocated compress object will be freed and bad_compress will be
returned, so error messages written to the allocated compress struct
will be lost. This patch writes error messages to bad_compress for error
paths in compress_open, ensuring that error messages are returned
correctly.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
git://git.alsa-project.org / tinycompress.git / log