Author: Eugene Nalimov
Date: 19:23:47 05/09/04
Go up one level in this thread
On May 09, 2004 at 03:58:05, Dieter Buerssner wrote:
>I tried to update to newer Nalimov code (with support of the split TBs). Seemed
>all easy, but then the program crashed. I had downloaded krppkr and kqppkq to
>test. For some reason, I forgot to download the kqqpkq.0.nbb/nbw.emd files.
>After the crash, I tried to check integrity with Wilhelm. It did not complain
>about the missing files - just said installed/integer.
>
>In the actual Nalimov code, one error condition seems to be missing.
>
>/* Probe TB - lower level (not exportable) function */
>
>static int TB_FASTCALL TbtProbeTable(int iTb, color side, unsigned indChunk,
> unsigned indInChunk)
>{
> [...]
> /* Now read it from the disk */
> FILE *fp;
> size_t cb;
>
>/* First, check: is necessary file opened?
> As files are not thread-safe, lock file */
> Lock(ptbd->m_rglockFiles[side]);
> if (ptbd->m_fSplit) {
> iExtent = indChunk >> (31 - LOG2_TB_CB_CACHE_CHUNK);
> iPhysicalChunk = indChunk - (iExtent << (31 - LOG2_TB_CB_CACHE_CHUNK));
> } else {
> iExtent = 0;
> iPhysicalChunk = indChunk;
> }
> fp = ptbd->m_rgfpFiles[side][iExtent];
> if (NULL == fp) {
>/* Not - try to open it */
> pszFileName = ptbd->m_rgpchFileName[side][iExtent];
> if (NULL != pszFileName) {
> [carefully tries to open the file with all sort of error checks]
> }
>
> [But an else case is missing, and later code assumes, that fp is initialized
>correctly. At least this is my understanding of the code. ]
> }
> [Now using fp, which might be NULL]
>
>Regards,
>Dieter
Add "else goto ERROR_LABEL;". I believe that is enough.
Will add it to the new code I Bob will release soon.
Thanks,
Eugene
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.