how to compile with nginx?

[osevan]

I want compile with nginx source, but not with .so or .a file i want compile buildin directly in nginx source to get maximum performance.

I dont want configure --with-ld-opt="-lsnmalloc"

I want directly embedded inside nginx compile toolchain. How i can do this?

Thanks and

best regards

[mjp41]

I haven't tried, but I would start by adding the .a but built with -DSNMALLOC_IPO=On. This should build a LTO/IPO enabled include, and then you can try that. This is not an option that is in the test harness, and I don't personally use it, so it might or might not work.

An alternative would be to add src/snmalloc/override/malloc.cc

to the build, and enable LTO. If it is a C++ project, then you might want to override new/delete. So you can include:

src/snmalloc/override/new.cc

instead of the malloc.cc unit. The new.cc includes malloc.cc already.

Please, let us know how you get on.

[osevan]

Could you try make patch for nginx

And we could point snmalloc sources at configure level to solve this

[mjp41]

I suggest you benchmark with LD_PRELOAD, and if that demonstrates that there is a win in using snmalloc. Removing the PLT indirection will improve performance, but if the different allocators you try are all going via a PLT indirection then it should be a relatively fair comparison to then decide if you want to do additional work to remove the PLT indirection.

[osevan]

Snmalloc will be nice .

I know that very well

[mjp41]

So I quickly tried the IPO option it seems to work, but doesn't give any inline above using the .a in the application I used.

 cmake -DSNMALLOC_IPO=On .. -DCMAKE_CXX_COMPILER=clang++ -GNinja -DSNMALLOC_STATIC_LIBRARY_PREFIX=""

will build

libsnmallocshim-static.a

which you can add to your build like

clang++ -flto test.cc -O3 libsnmallocshim-static.a -O test

But I don't see any advantage in the disassembly over not using IPO.