On Thu, Feb 21, 2013 at 2:35 PM, Kai Tietz <[email protected]> wrote: > Well, I am right now playing with your test-case. I think we should > approach it differently.
How? > As by your code the TLS-region contains > always one element. "As by your code" - you mean as by the current tlssup.c implementation? As long as it places something at the start of the TLS section _tls_start, I don't see a way to avoid keeping that thing around while keeping the SECREL offsets valid. > I just noticed that in pe.sc some weird position > for .tls$AAA vs .tls$ZZZ is used. What is weird about them? The only peculiarity I see is that .tls$ZZZ is already picked up by *(SORT(.tls$*)) (and thus compilers have to take care to emit only TLS section names that are actually sorted before .tls$ZZZ), but maybe I'm missing something. > I am right now on travel, so I will continue to look further into this > at weekend. Of course, it would be great if you could come up with a better approach, but for my part, I'm not unsatisfied with the current situation. It seems to match what MSVC does (initial values differ from the ones in my previous asm example): ----- .tls:00404000 ; Section 4. (virtual address 00004000) .tls:00404000 ; Virtual size : 0000000D ( 13.) .tls:00404000 ; Section size in file : 00000200 ( 512.) .tls:00404000 ; Offset to raw data for section: 00001600 .tls:00404000 ; Flags C0000040: Data Readable Writable .tls:00404000 ; Alignment : default .tls:00404000 ; ============================================= .tls:00404000 .tls:00404000 ; Segment type: Pure data .tls:00404000 ; Segment permissions: Read/Write .tls:00404000 _tls segment para public 'DATA' use32 .tls:00404000 assume cs:_tls .tls:00404000 ;org 404000h .tls:00404000 __tls_start db 0 .tls:00404001 db 0 .tls:00404002 db 0 .tls:00404003 db 0 .tls:00404004 __tlsVar2 db 59h ; Y .tls:00404005 db 0 .tls:00404006 db 0 .tls:00404007 db 0 .tls:00404008 __tlsVar db 2Ah ; * .tls:00404009 db 0 .tls:0040400A db 0 .tls:0040400B db 0 .tls:0040400C __tls_end db 0 ----- David ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_d2d_feb _______________________________________________ Mingw-w64-public mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
