Re: [PATCH] full device

[Mikulas Patocka <mikulas artax karlin mff cuni cz>]

One more patch fixing two issues:

stat.st_blocks may be 64-bit, but Midnight commander truncates it to long

If the file has too many blocks, display overflows the right end of the 
panel (because there is no size_trunc_len when printing block count).

Mikulas

> Another patch:
>
> When the device is out of free inodes or blocks, MC will display "No node 
> information" or "No space information" --- actualy when the filesystem 
> declines to report node or space information, it sets total number of 
> blocks/nodes to 0. When it sets number of free inodes/blocks 0, it means that 
> it is full.
>
> Mikulas
>
> > Hi
> >
> > I've found on three machines (I wonder that no one noticed it so far) that 
> > mc displays incorrect information on inode count, for example:
> >
> > |  Filesystem: /                       |
> > |  Device:    /dev/md0                 |
> > |  Type:      ext2                     |
> > |  Free space: 161G (0%) of 552G       |
> > |  Free nodes: 32864471 (-27%) of 364380
> >
> > |  Filesystem: /home                   |
> > |  Device:    /dev/md0                 |
> > |  Type:      xfs                      |
> > |  Free space: 15G (13%) of 117G       |
> > |  Free nodes: 64702776 (-31%) of 669767
> >
> > |  Filesystem: COMMON:/                |
> > |  Device:    DISK1-P2:/               |
> > |  Type:      SPADFS                   |
> > |  Free space: 36G (40%) of 88G        |
> > |  Free nodes: 74794575 (-5%) of 1850848
> >
> > --- the bug is obviously caused by integer overflow when multiplying by 
> > 100. When I further investigated the code, I found that it stores free 
> > space and free inode count only in "int" variable --- it will cause 
> > problems with devices with more than 2^31 blocks (they already exists, but 
> > I don't have such). --- so I fixed both block and inode display to use 
> > double instead --- after aplying the patch, the output looks sane:
> >
> > |  Filesystem: HOME:/                  |
> > |  Device:    DISK1-P2:/               |
> > |  Type:      SPADFS                   |
> > |  Free space: 36G (40%) of 88G        |
> > |  Free nodes: 71M (40%) of 177M       |
> >
> > --- maybe you could change double to "long long" but I'm not sure if it 
> > exists on all machines --- a configure test would probably be needed.
> >
> > Mikulas
--- /tmp/info.c	2006-12-29 02:59:50.000000000 +0200
+++ src/info.c	2006-12-29 03:02:01.000000000 +0200
@@ -165,8 +165,9 @@
 	    size_trunc_len(buffer, 9, st.st_size, 0);
 	    printw (const_cast(char *, _("Size:      %s")), buffer);
 #ifdef HAVE_STRUCT_STAT_ST_BLOCKS
+	    size_trunc_len(buffer, 6, st.st_blocks, 0);
 	    printw (const_cast(char *, (st.st_blocks==1) ?
-		_(" (%ld block)") : _(" (%ld blocks)")), (long) st.st_blocks);
+		_(" (%s block)") : _(" (%s blocks)")), buffer);
 #endif
 	}