On Mon, Jun 18, 2007 at 10:26:03PM -0700, Joe Wilson wrote:
> It is sort-of a NestedVM bug in the sense that it did not have
> a proper declaration of ftruncate(int fd, off_t length) which
ftruncate is declared in sys/unistd.h as:
ftruncate(int fd, off_t length)
which is correct.
> ultimately causes the mips cross-compiler to push a 64 bit value
> for the off_t parameter instead of the 32 bit value NestedVM/mips
> was expecting. This resulted in a psuedo-random stack value for
Why would the compiler pass a 64-bit value? off_t is a long (a "c"
long, which is 32-bits under NestedVM, not a java long).
> --- src/os_unix.c 8 Jun 2007 18:27:03 -0000 1.133
> +++ src/os_unix.c 19 Jun 2007 05:04:52 -0000
> @@ -1271,7 +1271,7 @@ int sqlite3UnixSyncDirectory(const char
> static int unixTruncate(OsFile *id, i64 nByte){
> int rc;
> assert( id );
> - rc = ftruncate(((unixFile*)id)->h, nByte);
> + rc = ftruncate(((unixFile*)id)->h, (off_t)nByte);
Ahh... I'd argue this is a clear bug in sqllite. It is assuming off_t
is the same as i64 (which is presumably a typedef to long long).
However, c's implicit narrowing conversions should convert the i64 to
off_t by taking the lower 32-bits, which should still be correct as
long as your offsets fit into 32-bits.
Unless of course you don't have a type declaration for ftruncate in
scope, in which case it will screw up.
-Brian
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"SQLiteJDBC" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/sqlitejdbc?hl=en
-~----------~----~----~----~------~----~------~--~---
