2009-04-21 02:40:28     fseek

2009-04-21 02:40:28     fseek

nive n (INDIA)

Message: 72922   

Hi,

I have a issue with fseek function.

ret  = fseek (fileptr,seek_offset,SEEK_CURR)

fseek is called with offset : 2012100347. The seek offset is an invalid one and is beyond the file size. fseek should return -1 in this case. But for this case , fseek returns success for more than 4000 times and finally error is returned.

########################################

seek offset : 2012100347, Return Value : 0

seek offset : 2012100347, Return Value : 0

.........

.........

.........

seek offset : 2012100347, Return Value : -1

#########################################

Pls let me know how to resolve this?

QuoteReplyEditDelete

2009-04-21 07:48:48     Re: fseek

Robin Getz (UNITED STATES)

Message: 72976   

Nive:

If you compile the same application on your desktop - does it work or fail?

What toolchain release are you using?

-Robin

QuoteReplyEditDelete

2009-04-21 08:09:01     Re: fseek

nive n (INDIA)

Message: 72977   

Hi,

I am using 2008r1.5-14 toolchain.

It is working fine in VC++.

Regards,

nive

QuoteReplyEditDelete

2009-04-21 08:26:14     Re: fseek

Robin Getz (UNITED STATES)

Message: 72981   

Nive:

Does it work on your Linux host with gcc and glibc?

-Robin

QuoteReplyEditDelete

2009-04-22 00:27:51     Re: fseek

nive n (INDIA)

Message: 73037   

Hi,

I  tried on Linux host with gcc and glibc.

There too i am facing the same issue.

Thanks,

Nive

QuoteReplyEditDelete

2009-04-22 05:30:21     Re: fseek

Yi Li (CHINA)

Message: 73054   

Nive,

I think this is the expected behavior. In manual page of lseek I read:

The lseek() function allows the file offset to be set beyond the end of the file (but this does not change the size  of

       the  file).   If  data  is  later written at this point, subsequent reads of the data in the gap (a "hole") return null

       bytes ('\0') until data is actually written into the gap.

-Yi

QuoteReplyEditDelete

2009-04-22 08:10:05     Re: fseek

Robin Getz (UNITED STATES)

Message: 73058   

Nive:

Yeah - Yi is correct - if you get that result with gcc & glibc - that is the expected behaviour and gcc/Blackfin/uClibc are operating 100% correct.

-Robin

QuoteReplyEditDelete

2009-04-22 08:14:26     Re: fseek

nive n (INDIA)

Message: 73059   

Thanks for your info.