Feature/openssl 1.0.1k

[indutny]

cc @bnoordhuis

[indutny]

Good news, they have finally fixed the EVP message!

[indutny]

Another good news: keypress patch should be no longer floated, because it is included in 1.0.1k!

[indutny]

@bnoordhuis: please check that your x32 changes are still in. I have used opensslconf from the origin/v1.x, so I assume - yes.

[bnoordhuis]

Can you make the CI check this PR? I suspect it's going to break the Windows build because of the removal of (at least) x86_64-win32-masm.asm.

[indutny]

@bnoordhuis : https://jenkins-iojs.nodesource.com/job/iojs+any-pr+multi/48/

[indutny]

@bnoordhuis: appears to be building fine on windows.

[bnoordhuis]

@indutny gyp is definitely complaining:

Warning: Missing input files: deps\openssl\openssl\crypto\bn\asm\x86_64-win32-masm.asm

The fact that it builds is presumably because the Windows bots build 32 bits binaries only. Maybe @rvagg can confirm.

[bnoordhuis]

This PR also seems to break parallel/test-https-foafssl on Windows.

[indutny]

@bnoordhuis: pushed update for GYP.

[indutny]

@bnoordhuis : regarding foafssl - it is not completely evident to me that this is caused by this PR.

[piscisaureus]

@indutny Would you mind also updating https://github.com/iojs/io.js/blob/v1.x/deps/openssl/asm/Makefile? I noticed some asm files er axed in this release.

[piscisaureus]

Building x64 on windows:

openssl.lib(bn_word.obj) : error LNK2001: unresolved external symbol bn_mul_wor
ds [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_word.obj) : error LNK2001: unresolved external symbol bn_div_wor
ds [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_add.obj) : error LNK2001: unresolved external symbol bn_add_word
s [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_mont.obj) : error LNK2001: unresolved external symbol bn_mul_add
_words [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_mont.obj) : error LNK2001: unresolved external symbol bn_sub_wor
ds [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_mul.obj) : error LNK2001: unresolved external symbol bn_mul_comb
a8 [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_mul.obj) : error LNK2001: unresolved external symbol bn_mul_comb
a4 [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_sqr.obj) : error LNK2001: unresolved external symbol bn_sqr_comb
a4 [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_sqr.obj) : error LNK2001: unresolved external symbol bn_sqr_comb
a8 [D:\iojs\deps\openssl\openssl-cli.vcxproj]
openssl.lib(bn_sqr.obj) : error LNK2001: unresolved external symbol bn_sqr_word
s [D:\iojs\deps\openssl\openssl-cli.vcxproj]
D:\iojs\Release\openssl-cli.exe : fatal error LNK1120: 10 unresolved externals
[D:\iojs\deps\openssl\openssl-cli.vcxproj]

[indutny]

@piscisaureus could you please try it again with a new commit?

[piscisaureus]

Trying

[piscisaureus]

@indutny yes, that works

WRT that missing file - I wrote that myself one day when I was really bored: 3568edf. I don't mind axing it; since openssl now seems properly maintained I no longer feel the need to outsmart them.

[indutny]

@piscisaureus do you want me to revert it?

[piscisaureus]

@indutny no, it works. That asm file I created was derived from the gcc asm implementation, but since that file has been updated since it seems safer to exclude my own version. Go ahead and land.

[indutny]

Ok, any objections @bnoordhuis ?

[bnoordhuis]

Well, I still don't get why parallel/test-https-foafssl is failing now when it was passing before.

[indutny]

@bnoordhuis let's give it another spin on jenkins? https://jenkins-iojs.nodesource.com/job/iojs+any-pr+multi/49/

[indutny]

@bnoordhuis seems to be timeout-ing anyway. @piscisaureus does it terminate when you run it outside of the test suite? (./iojs test/parallel/test-https-foafssl.js)

[piscisaureus]

@indutny

That'd be Release\iojs.exe test/parallel/test-https-foafssl.js and yes, it hangs.

[rvagg]

brave enough to land this prior to 1.0.0 or should we hold off?

[bnoordhuis]

@piscisaureus You mean it hangs with or without this PR applied? If it's without, then the failure is not a regression and this PR should be good to go.

[indutny]

Going to see it myself.

[indutny]

Stupid me, I thought that the patch was included, but it was the artifact of cherry-pick which has disappeared after cherry-pick --abort. Here is the latest build: https://jenkins-iojs.nodesource.com/job/iojs+any-pr+multi/53/

[indutny]

jenkins is good with it now. @bnoordhuis could you please land it if it LGTY?

[bnoordhuis]

I carefully scrutinized each and every one of those +1,984 −5,363 changed lines. LGTM!

[indutny]

Landed in eebdf7a, b910613, ced41b0, a76811c, 7c4a50d ! Thanks everyone!