[squid-dev] squid-3.5.7 -- xstrerror missing

Linda A. Walsh squid-user at tlinx.org
Wed Aug 12 02:17:41 UTC 2015


Linda A. Walsh wrote:
>>
>> I dont see anywhere on this list:
>>  --disable-inline        to prevent inline optimizations
>>  --disable-optimizations to prevent compiler -O levels
>>  --disable-arch-native   to prevent -march=native auto-detection
>>
>> You said you disabled them ?  
----
Now I see the confusion... I had a bunch of option-by-option
optimizations that I've used in the past  most of them building on -O2, like
these:
-O2 -m64 -fasynchronous-unwind-tables
-fbranch-target-load-optimize -fdelete-null-pointer-checks
-fgcse-after-reload -fgcse-las -fgcse-sm -fgraphite-identity
-finline-functions -fipa-pta -fivopts -floop-block -floop-flatten
-floop-interchange -floop-strip-mine -flto -fmessage-length=0
-fpredictive-commoning -frename-registers -freorder-blocks-and-partition
-ftracer -fsched-stalled-insns=1 -fsched-stalled-insns-dep=1
-ftree-loop-linear -ftree-loop-distribution
-ftree-loop-distribute-patterns -ftree-loop-im -ftree-loop-ivcanon
-ftree-vectorize -ftree-slp-vectorize -funswitch-loops -funwind-tables
-fvariable-expansion-in-unroller -fvect-cost-model -fweb -fwhole-program
-march=native -fpie -pipe -march=native  (all in CFLAGS -- they have
worked before on most programs)... Anyway, I meant I got rid of all
the specific switches and went to:

"CFLAGS="-O2 -m64 -march=native -fpie -pipe"
----
So it still has -O2 and -march=native -- but usually after I get
something working, I read over the individual items and look at ones
that look like they won't increase memory (much, if at all), and that
mostly take extra compile time -- even if they hurt 'debugging' -- unless
it is a program I'm doing active development on and need such..
Then I go for the debug heavy (-Og -g3 -ggdb type options) to get all the
help I can get.

Sorry for the confusion...




More information about the squid-dev mailing list