Christian Smith wrote:
Just run the test suite on NetBSD/sparc64, and test pagesize-2.3000.7 died
with a SIGBUS. Full stack trace at bottom of mail[0].
#0 0x0000000000108d80 in balance_nonroot (pPage=0x39dc10)
at ../sqlite/src/btree.c:3045
3045 p->aData = &((u8*)p)[-pBt->pageSize];
(gdb) print *p
$2 = {isInit = 70 'F', idxShift = 124 '|', nOverflow = 83 'S',
intKey = 68 'D', leaf = 113 'q', zeroData = 51 '3', leafData = 86 'V',
hasData = 46 '.', hdrOffset = 112 'p', childPtrSize = 118 'v',
maxLocal = 12650, minLocal = 18531, cellOffset = 29040, idxParent = 15423,
nFree = 26698, nCell = 22633, aOvfl = {{
pCell = 0x767a74214679514a <Address 0x767a74214679514a out of bounds>,
idx = 29743}, {pCell = 0x0, idx = 0}, {pCell = 0x0, idx = 0}, {
pCell = 0x0, idx = 0}, {pCell = 0x0, idx = 0}}, pBt = 0x0, aData = 0x0,
pgno = 0, pParent = 0x0}
What's does "print *pBt" say? How about "print (u8*)p"?
--
D. Richard Hipp -- [EMAIL PROTECTED] -- 704.948.4565