25/03/2021 19:50, Tyler Retzlaff: > On Thu, Mar 25, 2021 at 09:00:54AM +0100, Thomas Monjalon wrote: > > 25/03/2021 01:09, Tyler Retzlaff: > > > On Thu, Mar 25, 2021 at 01:41:47AM +0300, Dmitry Kozlyuk wrote: > > > > 2021-03-24 14:55 (UTC-0700), Tyler Retzlaff: > > > > > > > > Can't speak for Seastar and Click, but we never hit this particular > > > > issue. > > > > What is the minimal complete example to reproduce? > > > > > > interesting, i did a bit more digging and it is clear why. i'm consuming > > > headers and crt that have been enhanced to enforce c++ standards > > > compliance. after dumping the preprocessed output i suspect this cannot > > > be reproduced with the publicly available headers it seems i'm the only > > > "beneficiary" heh. > > > > > > so the minimal example is really just to #include <list> after including > > > say rte_windows.h but it won't reproduce for public users of the sdk. > > > > > > if there is a great deal of concern with respect to compatibility i > > > guess as the single consumer who has this problem i can just #undef asm > > > since we don't use inline asm anyway. > > > > > > thoughts? > > > > No please don't do this. > > There is an issue in DPDK which must be fixed. > > It seems your patch is correct, I am waiting for confirmation of others. > > thanks, i'm glad for the desire to make the proper fix. we'll wait for > others to comment.
Nobody dares to comment on the asm keyword, so I'll just take your patch with few more explanations.
