Hide Forgot
Description of problem: I just tried to compile package star-1.5a25-6 from Redhat Fedora Core 3 Test 1. The compiler said 1. header.c(1271): warning #175: subscript out of range The source code is if (ptb->dbuf.t_name[NAMSIZ] == '\0') { Suggest if (ptb->dbuf.t_name[NAMSIZ - 1] == '\0') { might be better code. 2. header.c(1283): warning #175: subscript out of range The source code is ptb->dbuf.t_name[NAMSIZ] = ' '; This will never work in a month of Sundays. Try ptb->dbuf.t_name[ NAMSIZ - 1] = ' '; 3. header.c(1372): warning #175: subscript out of range header.c(1373): warning #175: subscript out of range header.c(1374): warning #175: subscript out of range header.c(1375): warning #175: subscript out of range header.c(1376): warning #175: subscript out of range header.c(1377): warning #175: subscript out of range The source code is xname = ptb->dbuf.t_name[NAMSIZ]; ptb->dbuf.t_name[NAMSIZ] = '\0'; /* allow 100 chars in name */ xlink = ptb->dbuf.t_linkname[NAMSIZ]; ptb->dbuf.t_linkname[NAMSIZ] = '\0'; /* allow 100 chars in linkname */ xpfx = ptb->dbuf.t_prefix[PFXSIZ]; ptb->dbuf.t_prefix[PFXSIZ] = '\0'; /* allow 155 chars in prefix*/ Crikey, someone doesn't know much about exclusive upper limits on arrays in C. 4. header.c(1385): warning #175: subscript out of range header.c(1386): warning #175: subscript out of range header.c(1387): warning #175: subscript out of range The source code is ptb->dbuf.t_name[NAMSIZ] = xname; /* restore remembered value */ ptb->dbuf.t_linkname[NAMSIZ] = xlink; /* restore remembered value */ ptb->dbuf.t_prefix[PFXSIZ] = xpfx; /* restore remembered value */ More of the same. 5. tartest.c(249): warning #175: subscript out of range The source code is ptb->ustar_dbuf.t_name[100] == '\0') { 6. tartest.c(255): warning #175: subscript out of range Duplicate. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Try report to the upstream. I think it's a hack. Ugly, but it works.
>Try report to the upstream. Sorry, haven't got time to do this. I had rather hoped a member of the redhat team might fix it. Certainly, when I report similar bugs in other packages in redhat fedora Core 3 test 3, my bug reports get fixed. >I think it's a hack. Ugly, but it works. I think it's broken code, which needs fixing.
> Sorry, haven't got time to do this. > I had rather hoped a member of the redhat team might fix it. This is a poor excuse. If you truly care about this issue, then you would put the effort into reporting it properly, or at least beginning the discussion in the upstream project mailing list. On the other hand, if you can demonstrate that this causes a crash or security problem, then it quickly becomes a RH problem.
I've pointed Joerg at the bug report. He's a very active maintainer of his packages so I'm sure he'll be glad to get the report.
Joerg says "This is an extremely outdated version of star (more than a year). The current release is 1.5a52. The code is 100% correct - ignore the compiler warnings."
> The code is 100% correct - ignore the compiler warnings. This is not true for the RHEL3 code, or the latest "alpha" code I can download from ftp://ftp.berlios.de/pub/star ... the 1.5a52. The star.h file is... #define NAMSIZ 100 [...] struct star_header { char t_name[NAMSIZ]; /* 0 Dateiname */ [...] typedef union hblock { [...] struct star_header dbuf; [...] } TCB; ...the header.c file is... if (ptb->dbuf.t_name[NAMSIZ] == '\0') { [...] ptb->dbuf.t_name[NAMSIZ] = ' '; [...] xname = ptb->dbuf.t_name[NAMSIZ]; ptb->dbuf.t_name[NAMSIZ] = '\0'; /* allow 100 chars in name */ ...even the comment shows that he knows it's wrong. ptb is a pointer to the TCB union, NAMSIZ is only ever defined to be 100 ... so he's not being "clever" there either. It's probably not exploitable.
I can confirm the version of star in Fedora Core 3 Test 3 still has these array index bugs.
Alan Cox writes: >I've pointed Joerg at the bug report. Thanks. I've tried emailing him, with no answer. Is his email address in the package out of date ? >He's a very active maintainer of >his packages so I'm sure he'll be glad to get the report. Interesting, however this code is still broken a year later. Version 1.5a68-1, so I suspect there have been quite a few versions since 1.5a25-6. Would the star package benefit from a movement into the extras area ?
The original code is 100% correct. With the proposed patch, the code would no longer be correct. If you like to make a bug report please make a bug report against the C-compiler as this compiler does not honor the /* LINTED */ comment: /* LINTED */ if (ptb->dbuf.t_name[NAMSIZ] == '\0') {