Re: [exim-dev] Summary: exim 4.50 hp-ux 11.23 (ipf) startup …

Page principale
Supprimer ce message
Répondre à ce message
Auteur: V. T. Mueller
Date:  
À: exim-dev
Sujet: Re: [exim-dev] Summary: exim 4.50 hp-ux 11.23 (ipf) startup problems
Hello,

Michael Haardt schrieb:
>>B.11.23 ia64
>>   C-ANSI-C              C.06.00        HP C/ANSI C Compiler
>>A working solution here is setting CFLAGS to "-Ae -AOe +Onolimit".
> What does -AOe do? My compiler does not offer that.


It's an ipf-only option. From man cc:

-AOa
-AOe         See corresponding -Aa or -Ae option.  In addition,
              allows the optimizer to aggressively exploit the
              assumption that the source code conforms to the ANSI
              programming language C or C++ standard.  At present,
              the effect is to make +Otype_safety=ansi the default.
              As new independently-controllable optimizations are
              developed that depend on the ANSI standards, the flags
              that enable those optimizations may also become the
              default under -AOa/-AOe.


> -Ae is default here,
> so I don't need to specify that. It is useful to enable -z, though,
> in order to catch null pointer references. How about that on Itanium?


Well, you may or may not add "-z". According to the manuals, there
is no difference between PA and ipf here.

> What happens with +O2 on Itanium?


That option is available, but "exim -bd" gets broken.

> I hope we can find a set of flags that allows building Exim for both
> architectures, old and new compiler. +O3 fails on the old compiler on
> HP-PA, so we had to go back to +O2 there, but I doubt you will notice
> the difference.


What version of cc does "+O3" fail with? I had three working test
compiles on PA-RISC using that option yesterday. I agree that
presumably an application like exim - where speed relies mainly on
i/o - will not substantially benefit from the differences between
two adjacent optimization levels, though.

[common compiler flags] As far as I can see, this will be quite
difficult to achieve. The compiler on ipf appears to be a completely
different product (different options, different behaviour, different
message numbering scheme,..).


Kind regards,
vt

P.S.:

+Otype_safety=[off|limited|ansi|strong]
           Enable [disable] aliasing across types.


                       off  The default.  Specifies that aliasing
                            can occur freely across types.  This is
                            a synonym to +Onoptrs_ansi and
                            +Onoptrs_strongly_typed options in
                            legacy cc.


                       limited
                            Code follows ANSI aliasing rules.
                            Unnamed objects should be treated as if
                            they had an unknown type.


                       ansi Code follows ANSI aliasing rules, and
                            unnamed objects should be treated the
                            same as named objects.  This option is
                            synonym to +Optrs_ansi option in legacy
                            cc.


                       strong
                            Code follows ANSI aliasing rules, except
                            that accesses through lvalues of a
                            character type are not permitted to
                            touch objects of other types. This
                            assumes that field addresses are not
                            taken.  This option is synonym to
                            +Optrs_strongly_typed option in legacy
                            cc.


P.P.S.: Maybe we should continue off-list? I don't want to annoy
anyone with more details on our platform-specific issues...