On 21/7/2022 3:21 pm, Sebastian Huber wrote: > On 20.07.22 22:52, Chris Johns wrote: >> On 20/7/2022 6:20 pm, Sebastian Huber wrote: >>> On 20/07/2022 10:11, Chris Johns wrote: >>>>>> Are these TLS allocations in newlib based on a single use of a newlib >>>>>> call >>>>>> that >>>>>> brings in the reent stuff? How does that work? >>>>> No, each TLS object is defined in the translation unit which uses the >>>>> object. >>>>> The goal is to get only those objects linked in which are needed. >>>>> Currently, the >>>>> Newlib thread delete hook pulls in some objects, however, this dependency >>>>> will >>>>> go away once the TLS destructors are implemented. >>>> If I have task A and it opens a file and writes something to a file system >>>> will >>>> it pull in the TLS support for files? >>> >>> Yes. >>> >>>> Given the TLS support is static does this mean all threads in the A app >>>> allocate >>>> the newlib file TLS variables even if they never use newlib? >>> >>> Yes. >>> >>>> >>>> My understanding of the existing reent support is tasks that do not use >>>> newlib >>>> it never allocate a reent struct? >>> >>> No, the struct _reent is allocated for all tasks if you don't use >>> CONFIGURE_DISABLE_NEWLIB_REENTRANCY (which is not recommended). How would >>> you >>> know if a task doesn't use Newlib? >> >> I thought the get would do a check and then allocate on first use. It has >> been >> many years since I looked at the implementation. >> >> I think this change is a good one because it will reduce the needed memory to >> the amount used and it will have better performance. >> >> I think option you post is a good thing to add, maybe as >> --enable-newlib-reent-tls. What will be the default? > > You mean a proper sb-set-builder option and not just a --with-newlib-tls?
Not an sb option rather something using the --wtih/--without flags. I just renamed yours as `thread-local` may not be as well know as TLS. > I > would make it the default for the targets with known to work TLS support: > aarch64, arm, nios2, powerpc, riscv, and sparc. Sure it can be the default for the archs that can support it but a --without-newlib-tls option might be useful to someone using libdl if this change breaks their usage, ie linking from archives on the target. Chris _______________________________________________ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel
