[Bug 32461] New: Evolve client (.NET 4.0 app) crashes on startup due to broken libgcrypt AES-NI support (misaligned stack for SSE operations, upstream)
http://bugs.winehq.org/show_bug.cgi?id=32461 Bug #: 32461 Summary: Evolve client (.NET 4.0 app) crashes on startup due to broken libgcrypt AES-NI support (misaligned stack for SSE operations, upstream) Product: Wine Version: 1.5.19 Platform: x86 OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: -unknown AssignedTo: wine-bugs(a)winehq.org ReportedBy: focht(a)gmx.net Classification: Unclassified Hello folks, continuation of bug 32408 Adding this bug for documentation purposes. Prerequisite: 'winetricks -q dotnet40' (32-bit WINEPREFIX) Optional (the client will download/install on its own): 'winetricks -q msxml6' The app crashes in libgcrypt. With proper debug symbols for i686 libgcrypt we get: --- snip --- $ winedbg --gdb ./Updater.exe Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 62] 0x7e29453d in do_aesni_enc_aligned ( a=0x7e2cb398 "\001K\257\"x\246\235\063\035Q\200\020\066C\351\232gC\303\321Q\232\264\362͚x\253\t\245\021\275]\036\362\r\316ּ\274\022\023\032\307\305G\210\252\b\016\225\027\353\026wq\232\317r\200\206\004", <incomplete sequence \343>, b=0x577d3a4 "\360R'~o\004", ctx=0x577d1a8) at rijndael.c:710 710 asm volatile ("movdqu %[src], %%xmm0\n\t" /* xmm0 := *a */ Wine-gdb> bt #0 0x7e29453d in do_aesni_enc_aligned ( a=0x7e2cb398 "\001K\257\"x\246\235\063\035Q\200\020\066C\351\232gC\303\321Q\232\264\362͚x\253\t\245\021\275]\036\362\r\316ּ\274\022\023\032\307\305G\210\252\b\016\225\027\353\026wq\232\317r\200\206\004", <incomplete sequence \343>, b=0x577d3a4 "\360R'~o\004", ctx=0x577d1a8) at rijndael.c:710 #1 do_aesni (ctx=0x577d1a8, decrypt_flag=0, bx=0x577d3a4 "\360R'~o\004", ax=0x7e2cb398 "\001K\257\"x\246\235\063\035Q\200\020\066C\351\232gC\303\321Q\232\264\362͚x\253\t\245\021\275]\036\362\r\316ּ\274\022\023\032\307\305G\210\252\b\016\225\027\353\026wq\232\317r\200\206\004", <incomplete sequence \343>) at rijndael.c:1132 #2 0x7e294946 in rijndael_encrypt (context=0x577d1a8, b=0x577d3a4 "\360R'~o\004", a=0x7e2cb398 "\001K\257\"x\246\235\063\035Q\200\020\066C\351\232gC\303\321Q\232\264\362͚x\253\t\245\021\275]\036\362\r\316ּ\274\022\023\032\307\305G\210\252\b\016\225\027\353\026wq\232\317r\200\206\004", <incomplete sequence \343>) at rijndael.c:1155 #3 0x7e294f58 in selftest_basic_128 () at rijndael.c:1660 #4 0x7e294aff in selftest () at rijndael.c:1749 #5 do_setkey (keylen=16, key=0x7c906678 "!\312{\226\200j\356\030\247\256S\241\344\202", <incomplete sequence \304>, ctx=0x7c908670) at rijndael.c:209 #6 rijndael_setkey (context=0x7c908670, key=0x7c906678 "!\312{\226\200j\356\030\247\256S\241\344\202", <incomplete sequence \304>, keylen=16) at rijndael.c:444 #7 0x7e278df8 in cipher_setkey (c=0x7c9085f0, key=<optimized out>, keylen=16) at cipher.c:896 #8 0x7e26e5b4 in gcry_cipher_setkey (hd=0x7c9085f0, key=0x7c906678, keylen=16) at visibility.c:521 #9 0x7e39743c in wrap_gcry_cipher_setkey (ctx=0x7c9085f0, key=0x7c906678, keysize=16) at cipher.c:115 #10 0x7e31e462 in _gnutls_cipher_init (handle=0x7c901c64, cipher=GNUTLS_CIPHER_AES_128_CBC, key=0x7c901c5c, iv=0x7c901c54) at gnutls_cipher_int.c:71 #11 0x7e3294f7 in _gnutls_init_record_state (params=0x7c901c38, read=1, state=0x7c901c4c) at gnutls_constate.c:299 #12 0x7e329ac8 in _gnutls_epoch_set_keys (session=0x7c901290, epoch=1) at gnutls_constate.c:431 #13 0x7e32a198 in _gnutls_write_connection_state_init (session=0x7c901290) at gnutls_constate.c:602 #14 0x7e31280d in _gnutls_send_handshake_final (session=0x7c901290, init=1) at gnutls_handshake.c:2888 #15 0x7e315e41 in _gnutls_handshake_common (session=0x7c901290) at gnutls_handshake.c:3121 #16 0x7e31781e in gnutls_handshake (session=0x7c901290) at gnutls_handshake.c:2690 #17 0x7e4c1a1a in schan_imp_handshake (session=0x7c901290) at /home/focht/projects/wine/wine-git/dlls/secur32/schannel_gnutls.c:162 #18 0x7e4bff6c in schan_InitializeSecurityContextW (phCredential=0x188ed0, phContext=0x54ad0b0, pszTargetName=0xfc01bc, fContextReq=524572, Reserved1=0, TargetDataRep=16, pInput=0x12da3bc, Reserved2=0, phNewContext=0x577d7f0, pOutput=0x12da3d0, pfContextAttr=0x11c3138, ptsExpiry=0x577d8e8) at /home/focht/projects/wine/wine-git/dlls/secur32/schannel.c:766 #19 0x7e4c9180 in InitializeSecurityContextW (phCredential=0x577d8f0, phContext=0x577d9f4, pszTargetName=0xfc01bc, fContextReq=524572, Reserved1=0, TargetDataRep=16, pInput=0x12da3bc, Reserved2=0, phNewContext=0x11c6bfc, pOutput=0x12da3d0, pfContextAttr=0x11c3138, ptsExpiry=0x577d8e8) at /home/focht/projects/wine/wine-git/dlls/secur32/wrapper.c:346 ... --- snip --- Source: http://git.gnupg.org/cgi-bin/gitweb.cgi?p=libgcrypt.git;a=blob;f=cipher/rijn... --- snip --- 1626 /* Run the self-tests for AES 128. Returns NULL on success. */ 1627 static const char* 1628 selftest_basic_128 (void) 1629 { 1630 RIJNDAEL_context ctx; 1631 unsigned char scratch[16]; 1632 ... 1672 1673 rijndael_setkey (&ctx, key_128, sizeof (key_128)); 1674 rijndael_encrypt (&ctx, scratch, plaintext_128); 1675 if (memcmp (scratch, ciphertext_128, sizeof (ciphertext_128))) 1676 return "AES-128 test encryption failed."; 1677 rijndael_decrypt (&ctx, scratch, scratch); 1678 if (memcmp (scratch, plaintext_128, sizeof (plaintext_128))) 1679 return "AES-128 test decryption failed."; 1680 1681 return NULL; 1682 } --- snip --- "RIJNDAEL_context ctx" was not properly 16-byte aligned on stack, hence the crash in encrypt using SSE instructions. Using big brother I found some reports about this problem: --- Gentoo "dev-libs/libgcrypt-1.5.0-r2 - aes-ni segfaults" buglink: https://bugs.gentoo.org/show_bug.cgi?id=442568 patch: http://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/dev-libs/libgcrypt/f... --- Debian "libgcrypt11: New 1.5.0 version segfaults with NSS/PAM LDAP" buglink: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=643336 --- Upstream libgcrypt buglink: https://bugs.g10code.com/gnupg/issue1452 patch: http://git.gnupg.org/cgi-bin/gitweb.cgi?p=libgcrypt.git;a=commitdiff;h=a9697... --- The "final" patch (using __attribute__ ((aligned (16))) on "RIJNDAEL_context" structure) did not fix the issue for me. At least on my Fedora 16, gcc version 4.6.3 20120306 (Red Hat 4.6.3-2) still misaligned context buffers, causing segfault. I added "-mpreferred-stack-boundary=2" to CFLAGS to force gcc to realign the stack in function prolog. Fixed prolog code: --- snip --- .text:08001EE0 selftest_basic_128: .text:08001EE0 push ebp .text:08001EE1 mov ebp, esp .text:08001EE3 lea esp, [esp-0Ch] .text:08001EE7 and esp, 0FFFFFFF0h .text:08001EEA lea esp, [esp-22Ch] ... --- snip --- I suspect there are various other distros still broken regarding this bug hence you need to build a fixed libgcrypt version on your own. Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email Do not reply to this email, post in Bugzilla using the above URL to reply. ------- You are receiving this mail because: ------- You are watching all bug changes.
http://bugs.winehq.org/show_bug.cgi?id=32461 Anastasius Focht <focht(a)gmx.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Keywords| |dotnet, download Status|NEW |RESOLVED URL| |https://www.evolvehq.com/cl | |ient Resolution| |UPSTREAM --- Comment #1 from Anastasius Focht <focht(a)gmx.net> 2012-12-16 11:49:29 CST --- Hello folks, marking this bug 'upstream'. Thank you for your attention. Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email Do not reply to this email, post in Bugzilla using the above URL to reply. ------- You are receiving this mail because: ------- You are watching all bug changes.
https://bugs.winehq.org/show_bug.cgi?id=32461 Austin English <austinenglish(a)gmail.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED --- Comment #2 from Austin English <austinenglish(a)gmail.com> --- Closing. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
https://bugs.winehq.org/show_bug.cgi?id=32461 Austin English <austinenglish(a)gmail.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|CLOSED |RESOLVED --- Comment #3 from Austin English <austinenglish(a)gmail.com> --- This was inadvertently caught up in my unclosed bugs filter. NOTOURBUG should only be closed when fixed upstream. Setting back to RESOLVED NOTOURBUG. Sorry for the spam. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
https://bugs.winehq.org/show_bug.cgi?id=32461 Anastasius Focht <focht(a)gmx.net> changed: What |Removed |Added ---------------------------------------------------------------------------- URL|https://www.evolvehq.com/cl |https://web.archive.org/web |ient |/20140729230757/https://ass | |ets.evolvehq.com/client/Evo | |lveSetup.exe -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
http://bugs.winehq.org/show_bug.cgi?id=32461 Hans Leidekker <hans@meelstraat.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED --- Comment #4 from Hans Leidekker <hans@meelstraat.net> --- (In reply to Austin English from comment #3)
This was inadvertently caught up in my unclosed bugs filter. NOTOURBUG should only be closed when fixed upstream.
Setting back to RESOLVED NOTOURBUG.
We can close this now. GnuTLS stopped using libgcrypt long ago. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
participants (3)
-
wine-bugs＠winehq.org -
WineHQ Bugzilla
-
WineHQ Bugzilla