Instead of supporting configuration of the field and scalar size independently,
both are now controlled by the availability of a 64x64->128 bit multiplication
(currently only through __int128). This is autodetected from the C code through
__SIZEOF_INT128__, but can be overridden using configure's
--with-test-override-wide-multiply, or by defining
USE_FORCE_WIDEMUL_{INT64,INT128} manually.
e7daa9b [build] Tweak JNI macro to warn instead of error for JNI not found. (Karl-Johan Alm)
5b22977 [build] Update ax_jni_include_dir.m4 macro to deal with recent versions of macOS (Karl-Johan Alm)
Pull request description:
Prior to this patch, this macro fails to find a working directory for the JNI headers, and results in compile failure when doing
```
./configure --enable-experimental --enable-module-ecdh --enable-jni
```
on more recent macOS versions. The relevant commit upstream is [here](http://git.savannah.gnu.org/gitweb/?p=autoconf-archive.git;a=commit;h=ab23d25b1f1ae544fffdaa0a94a794798695c672) from the [GNU.org page for the macro](https://www.gnu.org/software/autoconf-archive/ax_jni_include_dir.html). The original (identical to the version in this commit) is [here](http://git.savannah.gnu.org/gitweb/?p=autoconf-archive.git;a=blob_plain;f=m4/ax_jni_include_dir.m4).
The compile failure:
```Bash
$ make
gcc -I. -g -O2 -Wall -Wextra -Wno-unused-function -c src/gen_context.c -o gen_context.o
gcc gen_context.o -o gen_context
./gen_context
CC src/libsecp256k1_la-secp256k1.lo
CC src/java/libsecp256k1_jni_la-org_bitcoin_NativeSecp256k1.lo
In file included from src/java/org_bitcoin_NativeSecp256k1.c:4:
In file included from src/java/org_bitcoin_NativeSecp256k1.h:2:
/Library/Java/JavaVirtualMachines/jdk1.7.0_71.jdk/Contents/Headers/jni.h:45:10: fatal error:
'jni_md.h' file not found
#include "jni_md.h"
^
1 error generated.
make: *** [src/java/libsecp256k1_jni_la-org_bitcoin_NativeSecp256k1.lo] Error 1
```
Tree-SHA512: 9a8305b3ed67eba0de728f91cf435857a676ba10507ab8481a3c03b50e1ce0469a3d79e751d0a697018789f21e2aa48b7eccca4d225520a3863fcf23f1fd487a
The only reason OpenSSL 1.1 was not supported was the removal of direct
access to r and s in ECDSA_SIG. This commit adds a simplified version of
ECDSA_SIG_get0 for < 1.1 that can be used like ECDSA_SIG_get0 in >= 1.1
OpenSSL 1.1 makes ECDSA_SIG opaque and our tests need access
inside this object.
The comparison tests against OpenSSL aren't important for most
users, but the build failing is...
Squashed and rebased. Thanks to @theuni and @faizkhan00 for doing
the majority of work here! Also thanks to @btchip for help with debugging
and review.
I Noticed this on OSX with clang, though it likely happens elsewhere as well.
The result is disabled x86_64 asm.
Due to missing escaping, this $0 was interpreted as the function name
SECP_64BIT_ASM_CHECK, causing the compile-check to be broken on some compilers.
The actual check looked like this:
int main()
{
uint64_t a = 11, tmp;
__asm__ __volatile__("movq SECP_64BIT_ASM_CHECKx100000000,%1; mulq %%rsi" : "+a"(a) : "S"(tmp) : "cc", "%rdx");
return 0;
}
It seems even more odd that it compiled anywhere.
This vastly shrinks the size of the context required for signing on devices with
memory-mapped Flash.
Tables are generated by the new gen_context tool into a header.
Pieter Wuille
fanquake
Jonas Nick
Gregory Maxwell
Alexander Block
Karl-Johan Alm
Pieter Wuille
GreenAddress
Cory Fields
Thomas Daede
Luke Dashjr
kiwigb