| [hatari-devel] asm56000.ttp problem |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/hatari-devel Archives
]
- To: Hatari Development <hatari-devel@xxxxxxxxxxxxxxxxxxx>
- Subject: [hatari-devel] asm56000.ttp problem
- From: Christian Zietz <czietz@xxxxxxx>
- Date: Wed, 3 Oct 2018 13:47:25 +0200
- Autocrypt: addr=czietz@xxxxxxx; prefer-encrypt=mutual; keydata= xsDiBDdn2AURBADksdHVyN55nv0lx4qGx+GQMrbo7zs7lSkAfhkgmgqp84xUeUiWI/kj1on/ wxkmJ96Yzt0ktDbZYM0C9Z66M3rLfXE1vXALHhegeMuOy/tVWybcohRrhfB7tmANTESJOZke 0lZZ59DcIfFoqLYErb6qX8nLPYnOv6sFubxnhuF9QQCg/3GaIR1sVK9Xq+b4B9BtVxd7cHMD /i2hAEOX3WY3K7PNZJziYF54uBbGiVS88W41l1RARcaeogIZcAKpFH3on+Tf60fAC85MCp17 QIeP44hj4Cf46B+UTVhf3EFG4IOsLRxUonpt7dKO8txsKFN/OFsjlPOuDyg7XMpEWkTWZetm HC9/0pcApIXSDnggde4T8AX6nn/+A/4hBOhPxuvkV7Uw/ebLYwXrLo2vt9OvvC1VfeywNseq PIkFX/+n/+niBS+Cb2ess2SVQNKJ9vP5+vBxg5AMfQXqk1ONldGQ/ARHmL6+Iuo47mO51e7R i691hq13wHUvyKh1AN7fpKI2m3YW55XEQ+3iTMIZcqfjr6xYgG8GJTppdc0gQ2hyaXN0aWFu IFppZXR6IDxjemlldHpAZ214Lm5ldD7CYwQQEQIAIwIZAQIeAQIXgAUCVGD5IgcLCQgHAwIB BhUIAgkKCwQWAgMBAAoJEFLLl/ZtoCXKubQAoIHNaurSMQB8MHDoTk3B7WHk2ApoAJ0egA8q aNoVj0kU4+OjeGzFiSHMOs7BTQQ3Z9gFEAgA9kJXtwh/CBdyorrWqULzBej5UxE5T7bxbrlL OCDaAadWoxTpj0BV89AHxstDqZSt90xkhkn4DIO9ZekX1KHTUPj1WV/cdlJPPT2N286Z4VeS Wc39uK50T8X8dryDxUcwYc58yWb/Ffm7/ZFexwGq01uejaClcjrUGvC/RgBYK+X0iP1YTknb zSC0neSRBzZrM2w4DUUdD3yIsxx8Wy2O9vPJI8BD8KVbGI2Ou1WMuF040zT9fBdXQ6MdGGze MyEstSr/POGxKUAYEY18hKcKctaGxAMZyAcpesqVDNmWn6vQClCbAkbTCD1mpF1Bn5x8vYlL IhkmuquiXsNV6TILOwACAgf+JhucyZDzOWGht9e0U71kC2bxIOr4iz+ADd3sxS62okrocHXp B9zYDhmJ74BFfC7xMd9bwWNj7YR0yiUdOzY27OcXcEkVmhVBW6AqxuRAKfmYMvvnyR5z5OP6 vg2YSzgOmooc5vequa5YIjLmFkuRlglLiEgdW9gPBFtirNqxOtAqSxEcRrblSn8JBEU51Ii6 SVVuo1nXOP11g8rVO4YvEED89pHT4jgLZu4th1N+mDumNZlqyUIxZ4tQyw3X2OWvEbKWGn2j h0ZywaomUTpVA+wiwxndawP40oowFYT8LNeLtfZyq6xPpQmT2DaNhP4gdy3qkDfnmXkc2zFM YukXo8I/AwUYN2fYBVLLl/ZtoCXKEQKA3QCfTJstYzXurbt9ZnoTU3SFQQmG0/wAoNX91nWM nsS7JOepPAzOUoke4AIi
- Openpgp: preference=signencrypt
Thorsten Otto schrieb:
> I don't think that this can easily be done. __NHANDLES is also used to set up
> an array of struct __open_file __open_stat[]. If you change the compares to
> accept higher values, certain function will suddenly write beyond the end of
> that array.
And this is precisely what appens within the open() call in asm56000:
Given a file handle>=40 it will write past the end of __open_stat, as I
have verified with the Hatari debugger. The effects will of course vary
depending on which variables are overwritten.
> BTW., current mintlib still suffers from the same problem: the handle returned
> by GEMDOS is used as an index into that array (offset by 3, to take into
> account valid negative handles). Since other programs compiled with current
> versions of mintlib don't seem to have that problem
Well, in current mintlib, someone put a "if (__OPEN_INDEX(rv) <
__NHANDLES)" into open.c before accessing __open_stat. I assume that
similar checks were added to all other functions accessing __open_stat.
That explains why programs using current mintlib are not crashing with
Harari's GEMDOS HD.
Christian
--
Christian Zietz - CHZ-Soft - czietz@xxxxxxx
WWW: http://www.chzsoft.de/
PGP/GnuPG-Key-ID: 0x52CB97F66DA025CA / 0x6DA025CA