Ignore:
Timestamp:
Oct 3, 2008, 1:50:44 AM (12 years ago)
Author:
Bruno Cornec
Message:
  • Fix #252
  • Improve OCFS2 support (ocfs2 is a SANE_FORMAT)
File:
1 edited

Legend:

Unmodified
Added
Removed
  • branches/2.2.7/mondo/src/common/libmondo-fifo.c

    r1645 r2030  
    1 /* libmondo-fifo.c
     1/*
    22   $Id$
    3 
    4 
    5 04/17
    6 - replaced INTERNAL_TAPE_BLK_SIZE with bkpinfo->internal_tape_block_size
    7 
    8 04/04/2004
    9 - ps wax is now ps wwax
    10 
    11 09/10/2003
    12 - p75 is now p80 in outopening tape call
    13 - negotiate largest buffer possible when opening tape drive
    14 - fall back to dd if buffer fails
    15 
    16 09/09
    17 - better logging for tape users
    18 - working on try_hard_to_fwrite(), try_hard_to_fread()
    19 - replaced internal w/ EXTERNAL 'buffer' exe.
    20 
    21 08/30
    22 - tweaked error msgs in try_hard_to_fwrite()
    23 
    24 08/02
    25 - updated is_incoming_block_valid() to make it
    26   return end-of-tape if >300 flotsam blocks
    27 
    28 05/02
    29 - when logging tape errors, don't repeat self
    30 
    31 04/24
    32 - added lots of log_OS_error()'s and assert()'s
    33 
    34 04/22
    35 - copy_file_from_here_to_there() --- added a bit of fault tolerance;
    36   if write fails, retry a few times before reporting error
    37 
    38 04/07/2003
    39 - line 866 --- set block_size used by internal buffer to 32768;
    40   was INTERNAL_TAPE_BLK_SIZE/2
    41 
    42 10/01 - 11/30/2002
    43 - is_incoming_block_valid() --- always make
    44   checksums %65536, just in case int size is
    45   odd (GRRR, ArkLinux)
    46 - disabled rotor-related superfluous warnings
    47 - added INTERNAL_TAPE_BLK_SIZE
    48 - do_sem() now returns int explicitly
    49 - changed internal_block_size
    50 - some irregularities (e.g. bad 'type'-ing) found by Kylix; fixed by Hugo
    51 
    52 09/01 - 09/30
    53 - change 64k to TAPE_BLOCK_SIZE
    54 - added internal_block_size; set it to TAPE_BLOCK_SIZE*2
    55 - if data is flowing FROM tape TO hard disk then set the threshold to 10 (not 75)
    56 - lots of multitape-related fixes
    57 - finally caught & fixed the 'won't finish unzipping last bigfile' bug
    58 - added WIFEXITED() after waitpid(), to improve multi-tape support
    59 
    60 08/01 - 08/31
    61 - trying to catch & stop 'won't finish unzipping last bigfile' bug by
    62   changing the copy_file_rom_here_to_there code
    63 - changed % threshold from 95 back to 75
    64 - don't insist on full 256K write of last block to tape
    65 - if >25 secs go by & all data (AFAIK) has been copied thru by FIFO wrapper
    66   and g_tape_stream is _still_ not closed then shrug shoulders & pthread_exit
    67   anyway...
    68 - change fprintf()'s to log_it()'s
    69 - added a header+footer to each block as it is read/written to/from tape
    70   by copy_file_from_here_to_there
    71 - wrote workaround to allow >2GB of archives w/buffering
    72 - changed % threshold from 75 to 95
    73 - added calls to set_signals(); 'buffer' was killing mondoarchive as
    74   it terminated
    75 - cleaned up struct-passing, to improve reliability and eliminate
    76   some race conditions
    77 - changed some forks to pthreads
    78 - added some comments
    79 - created libfifo{.c,.h,-EXT.h}
    80 - copied core of 'buffer' here
    81 - added some other, Mondo-specific functions
    82 - hacked 'buffer' into user-friendliness
    833*/
    844
     
    9919#include <sys/stat.h>
    10020#include <sys/ipc.h>
    101 #include <sys/shm.h>
    102 #include <sys/sem.h>
    103 #include <sys/wait.h>
    104 #include <pthread.h>
    10521
    10622#include "my-stuff.h"
Note: See TracChangeset for help on using the changeset viewer.