aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Add a script that analyzes the source and prints module dependencies.lloyd2009-07-155-35/+109
| | | | | | | | | Useful for tracking where the big balls of mud are. Fix dependencies in gost_3411 (depends on the gost block cipher), and the TLS PRF (depends on HMAC). Also hide TLS_PRF::P_hash in an anonymous namespace instead of making it a private static function. I don't think this will affect binary compat, since it was statically linked.
* New module load_on directive 'never', to mark known-buggy codelloyd2009-07-154-2/+5
|
* Report modules which are not loaded, grouped by the reason why.lloyd2009-07-151-17/+29
|
* --with-build-dir fixeslloyd2009-07-141-7/+7
|
* merge of 'cd3b49c6cd609a1c81c7200e127063448f2254b7'lloyd2009-07-141-1/+2
|\ | | | | | | and 'e6d5e12d439e2f149f547f1afe47fea14e7863c2'
| * Generate the botan.doxy file in configure.pylloyd2009-07-131-1/+2
| |
* | Add support for a version suffix, so that botan reports its current versionlloyd2009-07-141-29/+32
| | | | | | | | | | | | | | | | as 1.8.5-pre instead of just 1.8.5. This suffix will also be reflected in the shared object soname. Should always be empty for final releases. Also add a --version flag to configure.py to match the one in configure.pl, which prints the current version and exits (same as perl version).
* | Also document Skein fixlloyd2009-07-141-0/+1
| |
* | Document that the test suite returns an error code nowlloyd2009-07-141-0/+1
| |
* | Avoid a crash in Skein_512::add_data if it is called with a zero-lengthlloyd2009-07-141-0/+3
| | | | | | | | input vector.
* | Canonicalize os names (from either uname or --os) against osinfo.aliases;lloyd2009-07-141-2/+12
| | | | | | | | | | noticed by Rickard Bondesson when he tried running configure.py on Solaris, aka SunOS.
* | Bump version numbers to 1.8.5-prelloyd2009-07-144-5/+8
| |
* | Avoid using optparse's append_const action, as it was only added in thelloyd2009-07-141-7/+21
|/ | | | | python 2.5 standard library. Replace it with a callback, which allows configure.py to run out of the box on python 2.4
* Return the test suite result code from main()lloyd2009-07-121-1/+1
|
* Return 1 to system if exception is caught in check proglloyd2009-07-121-0/+2
|
* Update configure.pl's version #1.8.4lloyd2009-07-121-1/+1
|
* Do a 1.8.4 release to fix MR nonce gen buglloyd2009-07-123-2/+5
|
* Fix nonce generation of Miller-Rabin testlloyd2009-07-121-1/+1
|
* Update readme and logfile for 1.8.3 release1.8.3lloyd2009-07-112-2/+2
|
* Remove unnecessary includes of <iostream>; either delete entirely as notlloyd2009-07-116-6/+4
| | | | needed, or replace with <iosfwd> or <ostream> where necessary.
* Fix generating primes between 4 and 7 bits. The problem was that whenlloyd2009-07-112-18/+23
| | | | | | | | verify mode is not set, by default the Miller-Rabin bases are chosen from the small primes. Generally speaking these make good test bases. However if the prime to be generated is very small, we will choose a base which is out of range. If the i'th prime is too big to be a base, then just choose a random integer of the appropriate size instead.
* static_cast a double before returning it as a u32bit to avoid a warninglloyd2009-07-101-1/+1
| | | | with some older versions of gcc
* Document fix to random_primelloyd2009-07-101-0/+1
|
* Fix some installation problems: the pkg-config and botan-config fileslloyd2009-07-101-12/+15
| | | | | | | were not being set to the right path, and the pkg-config file was set with the wrong name. Also build.h was not being installed by the make install target.
* Devai Tamas pointed out on the mailing list that random_prime with bitslloyd2009-07-101-3/+3
| | | | | | set to 2, 3, or 4 was not returning a random prime due to reducing the rng output modulo 1 instead of mod 2 in choosing which prime of that size to return. Oops.
* Various small code cleanups and slightly better error checking for info fileslloyd2009-07-091-16/+22
|
* Fix symlink generation when combined with --with-build-dirlloyd2009-07-091-42/+48
|
* Deal with the case where platform.processor() returns the empty string,lloyd2009-07-091-15/+12
| | | | | | which it is documented to do in some situations. Use a list comprehension instead of an explicit loop in choose_modules_to_use
* Don't recommend SHA-1 for new applications.lloyd2009-07-081-30/+33
| | | | Also change some examples using SHA-1 to use SHA-256 instead.
* Someone commented that they were using Blowfish because "No particularlloyd2009-07-081-94/+93
| | | | | | | reason but it was used in the tutorial." - it should have occured to me to change this a while ago. Switch to Serpent instead of Blowfish, and also replace most uses of SHA-1 with SHA-256 since SHA-1 is pretty broke these days.
* Various small code cleanupslloyd2009-07-081-13/+14
|
* Use a single regexp combined with | instead of three distinct searcheslloyd2009-07-081-3/+3
|
* In configure.py, check against the gcc version and toggle -fpermissivelloyd2009-07-081-1/+29
| | | | if needed
* Also match GCC 4.4 for -fpermissive checkslloyd2009-07-081-1/+1
|
* Correct detection of GCC 2.95 for setting -fpermissive. This had beenlloyd2009-07-081-1/+1
| | | | | wrong for a while, which suggests perhaps that 2.95 has finally died out in the wild. Praise be.
* Some modules using asm were not marked with 'load_on asm_ok'; fixlloyd2009-07-073-3/+3
|
* Make default logging INFO rather than DEBUGlloyd2009-07-071-1/+1
|
* CPU-specific engines are now only loaded if something depends on them,lloyd2009-07-079-11/+9
| | | | | | | | | | | | and all CPU-specific implementations now depend on the appropriate engine module. The most common problem before with this was that the SSE2 module was built, but the sole SSE2 code (SHA-1) was not (for instance, on an i686). This would cause a compile warning about the unused request object. Preventing unused engines from being built will also (very slightly) speed up the lookup process on most system.
* Cosmetic makefile generation stufflloyd2009-07-071-31/+57
| | | | | | | | | | Detect Jython; autodetection via platform does not work here, so tell the user they have to use --os and --cpu to set the target. Other than that Jython 2.5 seems to handle things nicely. Detect invalid settings for --with-endian during option parsing. Wrap some long lines for readability.
* More logging in configure.pylloyd2009-07-071-15/+24
|
* Fix out of tree buildslloyd2009-07-071-21/+20
|
* Clean up submodel guessing, and use the longest-match-first trick.lloyd2009-07-071-21/+21
| | | | Use Python's logging module instead of print for autoconfig output.
* If guess_processor() fails and just reutrns the base proc (platform.machine()),lloyd2009-07-071-1/+1
| | | | | first run that through canon_processor() to make sure we get a valid arch and submodel pairing.
* Form pkg-config output filename from version #s in build_configlloyd2009-07-031-8/+16
|
* Use /usr/bin/env to get to python, as it lives in /usr/local on some machineslloyd2009-07-021-1/+1
|
* Fix a subtle bug in the /dev/*random reader. The maximum ms wait time waslloyd2009-07-021-2/+3
| | | | | | | | set to 1000 ms (scaling based on amount of data requested). At 1000 ms exactly, we would form a timeval of 0 seconds and 1000000 usecs (ie, 1 second). Linux was fine with this, but FreeBSD 7.0's select was returning EINVAL. Fix things to properly create the timeval so that everyone is happy.
* Handle the mach_abi_linking flagslloyd2009-07-021-1/+15
|
* Add a bit of status output. Guess msvc as cc for Windows boxenlloyd2009-07-021-2/+11
|
* Add some aliases for 586/686 to match against what platform produces on WinXPlloyd2009-07-021-0/+3
|
* Document new configure.pylloyd2009-07-021-1/+1
|
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-20 04:38:18 +0000