 Roger Dingledine
|
5f051574d5
Happy new year!
|
18 years ago |
 Nick Mathewson
|
45757dafb7
Split PARANOIA_B into B1 and B2.
|
18 years ago |
|
Nick Mathewson
|
241310bbac
Split 0119_PARANOIA into 0119_PARANOIA_[ABC]. A is "this is suspicious, and we have not tried running without this yet". B is "this is suspicious, but the last time we tested, it was okay." C is "How could this possibly be the cause?"
|
18 years ago |
|
Nick Mathewson
|
55ac4f032c
Add a (diabled by default) option in crypto.h to disable most of the interesting crypto-related changes made on 0.1.1.9. This will help hunt bug 234.
|
18 years ago |
|
Nick Mathewson
|
1af630d32c
Bite the bullet and limit all our source lines to 80 characters, the way IBM intended.
|
18 years ago |
|
Nick Mathewson
|
e9b66ec906
Document CREATE_FAST better in the code. Move our key expansion algorithm into a separate function in crypto.c
|
18 years ago |
|
Roger Dingledine
|
485b2cb4dc
get rid of nick's crazy voodoo dh checking.
|
18 years ago |
|
Roger Dingledine
|
923ad87470
remove another unused function
|
18 years ago |
|
Nick Mathewson
|
9cec3a13f5
remove some functions that are not used; #if0 out some files that are not likely to be used.
|
18 years ago |
|
Roger Dingledine
|
8182eb69cd
a smidgen more hint when we fail to seed the rng
|
18 years ago |
|
Nick Mathewson
|
83d6b0387b
Try to build with OpenSSL 0.9.6. Lets pay attention to see if anybody complains.
|
18 years ago |
|
Nick Mathewson
|
887178e710
Oops. It looks like there *was* an easy way to make openssl do what we wanted.
|
18 years ago |
|
Roger Dingledine
|
5ef2927870
correct nick's comment
|
18 years ago |
|
Nick Mathewson
|
f14743a975
Note that much of check_dh_key is voodoo; make x in DH be only 320 bits for DH speed improvement: this wants auditing. [We have blessing from Ian on this.] (Note that DH in SSL is not yet affected.)
|
18 years ago |
|
Nick Mathewson
|
3c36a14ba6
Call ERR_remove_state() on the main thread on shutdown,too
|
18 years ago |
|
Nick Mathewson
|
9492424d3f
Per comments at the bottom of openssl/FAQ, call even more functions to
|
18 years ago |
|
Nick Mathewson
|
edf5698474
Start dividing log messages into logging domains. No, LD_ is not the best of identifiers. src/or has not been converted yet. Domains dont do anything yet.
|
18 years ago |
 Peter Palfrader
|
0d9aedfcea
Downgrade a few INFO level logs to DEBUG again. Also add two or three new
|
18 years ago |
|
Nick Mathewson
|
9bc2467626
Okay, try to use RAND_poll() from OpenSSL where available.
|
18 years ago |
|
Roger Dingledine
|
c4757e0705
LOG_ERR is for when we know we're going to exit. use LOG_WARN in other
|
18 years ago |
|
Nick Mathewson
|
150e5ac098
disable RAND_poll() for the alpha of the day; at least 24 hours of testing is in order for something like that.
|
18 years ago |
|
Nick Mathewson
|
a53ecc94f9
Add half our entropy from RAND_poll in OpenSSL. These know how to use egd (if present) openbsd weirdness (if present), vms/os2 weirdness (if we ever port there), and more in the future.
|
18 years ago |
|
Nick Mathewson
|
a89daaeca9
Once an hour (not just on startup) give OpenSSL some more entropy.
|
18 years ago |
|
Nick Mathewson
|
cc35e1720f
Using RAND_pseudo_bytes instead of RAND_bytes is an accident waiting to happen, and does not really speed us up much when we do it. So stop doing it.
|
18 years ago |
|
Nick Mathewson
|
ba24193ab5
Make doxygen marginally happier
|
18 years ago |
|
Nick Mathewson
|
f8c07e1f33
free EVP cipher information on shutdown to remove some spurious dmalloc complaints.
|
18 years ago |
|
Nick Mathewson
|
de198d800b
Never call free() on tor_malloc()d memory. This is unlikely to be our current leak, but it may help dmalloc work.
|
18 years ago |
|
Nick Mathewson
|
26e7a05725
even better function start checks; give dmalloc a chance of working.
|
18 years ago |
|
Nick Mathewson
|
92451f74a8
Reformat inconsistent function declarations.
|
18 years ago |
|
Nick Mathewson
|
5c53545d81
Add a bunch more warnings to out warning suite; resolve them; pack structs a little better.
|
18 years ago |
