Hi,
http://bugs.python.org/issue9116 was made a blocker this morning. I can neither verify the bug nor test the patch, so I would be grateful for any help, especially with patch review.
Also, the Windows 3.x buildbot shows failures compiling _ssl and _hashopenssl (for logs, see e.g. http://bit.ly/aSAnhh). I'd like to know if this is specific to that machine (I've already mailed David about it) or a general problem.
Thanks, Georg
Also, the Windows 3.x buildbot shows failures compiling _ssl and _hashopenssl (for logs, see e.g. http://bit.ly/aSAnhh). I'd like to know if this is specific to that machine (I've already mailed David about it) or a general problem.
This was a general problem - I forgot to update the external-common.bat for OpenSSL 1.0.0a. This is now fixed.
There is a related problem - apparently, OpenSSL now fails to build on AMD64. I look into this later today.
Regards, Martin
Am 31.07.2010 11:23, schrieb "Martin v. Löwis":
Also, the Windows 3.x buildbot shows failures compiling _ssl and _hashopenssl (for logs, see e.g. http://bit.ly/aSAnhh). I'd like to know if this is specific to that machine (I've already mailed David about it) or a general problem.
This was a general problem - I forgot to update the external-common.bat for OpenSSL 1.0.0a. This is now fixed.
There is a related problem - apparently, OpenSSL now fails to build on AMD64. I look into this later today.
Thanks Martin!
The buildbot still seems to have problems -- http://www.python.org/dev/buildbot/3.x.stable/builders/x86%20XP-4%203.x/buil... shows
cl /Fotmp32\o_dir.obj -Iinc32 -Itmp32 /MT /Ox /O2 /Ob2 -DOPENSSL_THREADS -DDSO_WIN32 -W3 -WX -Gs0 -GF -Gy -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -D_CRT_SECURE_NO_DEPRECATE -DOPENSSL_BN_ASM_PART_WORDS -DOPENSSL_IA32_SSE2 -DOPENSSL_BN_ASM_MONT -DSHA1_ASM -DSHA256_ASM -DSHA512_ASM -DMD5_ASM -DRMD160_ASM -DAES_ASM -DWHIRLPOOL_ASM -DOPENSSL_NO_IDEA -DOPENSSL_NO_RC5 -DOPENSSL_NO_MD2 -DOPENSSL_NO_MDC2 -DOPENSSL_NO_KRB5 -DOPENSSL_NO_JPAKE -DOPENSSL_NO_DYNAMIC_ENGINE /Zl /Zi /Fdtmp32/lib -c .\crypto\o_dir.c o_dir.c crypto\x86cpuid.pl win32n /MT /Ox /O2 /Ob2 -DOPENSSL_THREADS -DDSO_WIN32 -W3 -WX -Gs0 -GF -Gy -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -D_CRT_SECURE_NO_DEPRECATE -DOPENSSL_BN_ASM_PART_WORDS -DOPENSSL_IA32_SSE2 -DOPENSSL_BN_ASM_MONT -DSHA1_ASM -DSHA256_ASM -DSHA512_ASM -DMD5_ASM -DRMD160_ASM -DAES_ASM -DWHIRLPOOL_ASM -DOPENSSL_NO_IDEA -DOPENSSL_NO_RC5 -DOPENSSL_NO_MD2 -DOPENSSL_NO_MDC2 -DOPENSSL_NO_KRB5 -DOPENSSL_NO_JPAKE -DOPENSSL_NO_DYNAMIC_ENGINE >tmp32\x86cpuid.asm Access is denied. NMAKE : fatal error U1077: 'crypto\x86cpuid.pl' : return code '0x1' Stop.
Looks like it wants to execute a Perl script? That probably doesn't work that way on Windows...
Georg
crypto\x86cpuid.pl win32n /MT /Ox /O2 /Ob2 -DOPENSSL_THREADS -DDSO_WIN32 -W3 -WX -Gs0 -GF -Gy -nologo -DOPENSSL_SYSNAME_WIN32 -DWIN32_LEAN_AND_MEAN -DL_ENDIAN -D_CRT_SECURE_NO_DEPRECATE -DOPENSSL_BN_ASM_PART_WORDS -DOPENSSL_IA32_SSE2 -DOPENSSL_BN_ASM_MONT -DSHA1_ASM -DSHA256_ASM -DSHA512_ASM -DMD5_ASM -DRMD160_ASM -DAES_ASM -DWHIRLPOOL_ASM -DOPENSSL_NO_IDEA -DOPENSSL_NO_RC5 -DOPENSSL_NO_MD2 -DOPENSSL_NO_MDC2 -DOPENSSL_NO_KRB5 -DOPENSSL_NO_JPAKE -DOPENSSL_NO_DYNAMIC_ENGINE >tmp32\x86cpuid.asm Access is denied. NMAKE : fatal error U1077: 'crypto\x86cpuid.pl' : return code '0x1' Stop.
Looks like it wants to execute a Perl script? That probably doesn't work that way on Windows...
Unfortunately, it does... OpenSSL requires Perl to build it on Windows. And indeed, the build slave does have perl installed:
Found a working perl at 'c:\Perl\bin\perl.exe' Found an SSL directory at '....\openssl-1.0.0a'
However, since Christian Heimes last changed it, we are not passing PERL anymore into the makefile, so this broke.
In the past, we didn't need Perl on the slaves, anyway, since our external copy of OpenSSL had OpenSSL already pre-configured, and all Perl-generated files were included. In 1.0.0, OpenSSL started to generate assembler files with Perl, which broke the build. I have now added them to the external copy as well, so this is fixed.
Now, the XP build succeeded. For some reason, the Windows 7 build failed due to duplicate symbols. I have cleaned the build area, and it currently retries.
Regards, Martin
Le samedi 31 juillet 2010 à 11:23 +0200, "Martin v. Löwis" a écrit :
Also, the Windows 3.x buildbot shows failures compiling _ssl and _hashopenssl (for logs, see e.g. http://bit.ly/aSAnhh). I'd like to know if this is specific to that machine (I've already mailed David about it) or a general problem.
This was a general problem - I forgot to update the external-common.bat for OpenSSL 1.0.0a. This is now fixed.
Thanks Martin! Can http://bugs.python.org/issue8569 be closed now?
Regards
Antoine.
Am 31.07.2010 13:50, schrieb Antoine Pitrou:
Le samedi 31 juillet 2010 à 11:23 +0200, "Martin v. Löwis" a écrit :
Also, the Windows 3.x buildbot shows failures compiling _ssl and _hashopenssl (for logs, see e.g. http://bit.ly/aSAnhh). I'd like to know if this is specific to that machine (I've already mailed David about it) or a general problem.
This was a general problem - I forgot to update the external-common.bat for OpenSSL 1.0.0a. This is now fixed.
Thanks Martin! Can http://bugs.python.org/issue8569 be closed now?
I'm not sure whether there is consensus that this is not a security issue. I think it can be closed.
Regards, Martin
Le samedi 31 juillet 2010 à 13:58 +0200, "Martin v. Löwis" a écrit :
Am 31.07.2010 13:50, schrieb Antoine Pitrou:
Le samedi 31 juillet 2010 à 11:23 +0200, "Martin v. Löwis" a écrit :
Also, the Windows 3.x buildbot shows failures compiling _ssl and _hashopenssl (for logs, see e.g. http://bit.ly/aSAnhh). I'd like to know if this is specific to that machine (I've already mailed David about it) or a general problem.
This was a general problem - I forgot to update the external-common.bat for OpenSSL 1.0.0a. This is now fixed.
Thanks Martin! Can http://bugs.python.org/issue8569 be closed now?
I'm not sure whether there is consensus that this is not a security issue. I think it can be closed.
Well, regardless of whether it is a security issue, upgrading OpenSSL brings slight behaviour changes, so I don't think it can be done in the bugfix branches.
Regards
Antoine.
Am 31.07.2010 11:07, schrieb Georg Brandl:
Hi,
http://bugs.python.org/issue9116 was made a blocker this morning. I can neither verify the bug nor test the patch, so I would be grateful for any help, especially with patch review.
OK, since the problem the issue describes happens in a case where Python exits with a fatal error anyway, I've decided to defer this blocker.
Also, the Windows 3.x buildbot shows failures compiling _ssl and _hashopenssl (for logs, see e.g. http://bit.ly/aSAnhh). I'd like to know if this is specific to that machine (I've already mailed David about it) or a general problem.
This has been resolved now, so I'll go ahead and tag the current state for releasing.
Georg