Some clarifications about the Linux throughput problem

This commit is contained in:
Thomas Schmitt 2009-12-06 07:45:00 +00:00
parent 07f535c72b
commit 18b72f1602
2 changed files with 7 additions and 1 deletions

View File

@ -1 +1 @@
#define Cdrskin_timestamP "2009.12.06.073404" #define Cdrskin_timestamP "2009.12.06.074622"

View File

@ -1285,6 +1285,12 @@ int burn_disc_close_track_dvd_plus_r(struct burn_write_opts *o,
MAP_SHARED | MAP_ANONYMOUS, -1, (off_t) 0); MAP_SHARED | MAP_ANONYMOUS, -1, (off_t) 0);
is supposed to allocate a properly aligned buffer. is supposed to allocate a properly aligned buffer.
64 KB is supposed to be a safe size. 64 KB is supposed to be a safe size.
Actually mmap() seems to be the main cause for a positive effect
of O_DIRECT.
This simplified transmission function did not bring visible benefit.
So for now it is not worth to teach it all applicable details of old
CD sector oriented transmission.
@return 1= ok, go on , 2= no input with track->open_ended = nothing written @return 1= ok, go on , 2= no input with track->open_ended = nothing written
<= 0 = error <= 0 = error