Bogus rsync "Success" message when out of disk space

[John Van Essen]

JW,

(Ya know, I've been on this list for over a year and I still have
no idea what your first name is.  Anyway...)

I moved my comment above the loop.

Here's the (final) patch.  If you want to change the comment before
applying it, feel free to do so.

--- fileio.c.orig       Fri Jan 25 15:07:34 2002
+++ fileio.c    Sun Apr 27 12:29:17 2003
@@ -69,25 +69,28 @@
        return len;
 }
 
 
 
 int write_file(int f,char *buf,size_t len)
 {
        int ret = 0;
 
-       if (!sparse_files) {
-               return write(f,buf,len);
-       }
-
+       /* Normal writes are in this loop, too, so that retries
+          of partial writes will eventually set errno. */
        while (len>0) {
-               int len1 = MIN(len, SPARSE_WRITE_SIZE);
-               int r1 = write_sparse(f, buf, len1);
+               int r1;
+               if (sparse_files) {
+                       int len1 = MIN(len, SPARSE_WRITE_SIZE);
+                       r1 = write_sparse(f, buf, len1);
+               } else {
+                       r1 = write(f, buf, len);
+               }
                if (r1 <= 0) {
                        if (ret > 0) return ret;
                        return r1;
                }
                len -= r1;
                buf += r1;
                ret += r1;
        }
        return ret;
-- 
        John Van Essen <vanes002 at umn.edu>


On Sat, 26 Apr 2003, jw schultz <jw at pegasys.ws> wrote:
>On Sat, Apr 26, 2003 at 03:09:29PM -0500, John Van Essen wrote:
>> Patches welcome, eh, Paul?
>> 
>> Upon further (belated) investigation, there are 2 affected places
>> in receiver.c with this error message.  Both call write_file().
>> And write_file is called only in those two places.  So that is the
>> appropriate location to patch.  Especially since the obvious fix is
>> to use the rewrite code already there for the sparse file writes.
>
>This looks correct to me.  On partial write we need to
>retry.  The callers to write_file only test for
>completeness.
>
>I'd drop the added comments or put a generic
>"loop until all is written or error" at the top of the loop.
>The inline comments make the code harder to read.
>
>> -------------------------------------8<-------------------------------------
>> --- fileio.c.orig       Fri Jan 25 15:07:34 2002
>> +++ fileio.c    Sat Apr 26 12:16:25 2003
>> @@ -69,25 +69,28 @@
>>         return len;
>>  }
>>  
>>  
>>  
>>  int write_file(int f,char *buf,size_t len)
>>  {
>>         int ret = 0;
>>  
>> -       if (!sparse_files) {
>> -               return write(f,buf,len);
>> -       }
>> -
>>         while (len>0) {
>> -               int len1 = MIN(len, SPARSE_WRITE_SIZE);
>> -               int r1 = write_sparse(f, buf, len1);
>> +               int r1;
>> +               if (sparse_files) {
>> +                       int len1 = MIN(len, SPARSE_WRITE_SIZE);
>> +                       r1 = write_sparse(f, buf, len1);
>> +               } else {
>> +                       /* Normal writes are in this loop, too, so that */
>> +                       /*   retries of partial writes will set errno.  */
>> +                       r1 = write(f, buf, len);
>> +               }
>>                 if (r1 <= 0) {
>>                         if (ret > 0) return ret;
>>                         return r1;
>>                 }
>>                 len -= r1;
>>                 buf += r1;
>>                 ret += r1;
>>         }
>>         return ret;
>> -------------------------------------8<-------------------------------------
>> 
>> Test of unpatched rsync (bogus error message):
>> 
>> write failed on jvejunk2 : Success
>> rsync error: error in file IO (code 11) at receiver.c(243)
>> rsync: connection unexpectedly closed (200 bytes read so far)
>> rsync error: error in rsync protocol data stream (code 12) at io.c(165)
>> 
>> 
>> Test of patched rsync (correct error message):
>> 
>> write failed on jvejunk2 : No space left on device
>> rsync error: error in file IO (code 11) at receiver.c(243)
>> rsync: connection unexpectedly closed (200 bytes read so far)
>> rsync error: error in rsync protocol data stream (code 12) at io.c(165)
>> 
>> -- 
>>         John Van Essen <vanes002 at umn.edu>
>-- 
>________________________________________________________________
>        J.W. Schultz            Pegasystems Technologies
>        email address:          jw at pegasys.ws
>
>                Remember Cernan and Schmitt