Richard Henderson - Re: RFC: TLS improvements for IA32 and AMD64/EM64T (original) (raw)

This is the mail archive of the gcc-patches@gcc.gnu.orgmailing list for the GCC project.

• From: Richard Henderson
• To: Alexandre Oliva
• Cc: Mark Mitchell , gcc-patches at gcc dot gnu dot org
• Date: Thu, 22 Sep 2005 08:45:15 -0700
• Subject: Re: RFC: TLS improvements for IA32 and AMD64/EM64T
• References: ord5n91tx5.fsf@livre.oliva.athome.lsd.ic.unicamp.br <or4q8l1sv8.fsf@livre.oliva.athome.lsd.ic.unicamp.br> <20050916215356.GA26641@redhat.com> <or7jdfy5b6.fsf@livre.oliva.athome.lsd.ic.unicamp.br> <20050917180134.GA16067@redhat.com> <or8xxt5gal.fsf@livre.oliva.athome.lsd.ic.unicamp.br> <43301CD4.1070301@codesourcery.com> <or3bnx8u71.fsf@livre.oliva.athome.lsd.ic.unicamp.br>

On Thu, Sep 22, 2005 at 03:58:42AM -0300, Alexandre Oliva wrote:

•  && (!current_function_is_leaf || current_function_calls_alloca))

•  && (!current_function_is_leaf || current_function_calls_alloca

•  || (ix86_current_function_calls_tls_descriptor

•      && regs_ever_live[SP_REG])))

What's the point of checking regs_ever_live[SP_REG]?

•   rtx id = gen_rtx_SYMBOL_REF (SImode, "_TLS_MODULE_BASE_");

•   SYMBOL_REF_FLAGS (id)

•     |= TLS_MODEL_GLOBAL_DYNAMIC << SYMBOL_FLAG_TLS_SHIFT;

This should be created in one place, and cached.

r~

• Follow-Ups:
  • Re: RFC: TLS improvements for IA32 and AMD64/EM64T
    * From: Alexandre Oliva
• References:
  • Re: RFC: TLS improvements for IA32 and AMD64/EM64T
    * From: Alexandre Oliva
  • Re: RFC: TLS improvements for IA32 and AMD64/EM64T
    * From: Richard Henderson
  • Re: RFC: TLS improvements for IA32 and AMD64/EM64T
    * From: Alexandre Oliva
  • Re: RFC: TLS improvements for IA32 and AMD64/EM64T
    * From: Richard Henderson
  • Re: RFC: TLS improvements for IA32 and AMD64/EM64T
    * From: Alexandre Oliva
  • Re: RFC: TLS improvements for IA32 and AMD64/EM64T
    * From: Mark Mitchell
  • Re: RFC: TLS improvements for IA32 and AMD64/EM64T
    * From: Alexandre Oliva