On 5/2/19 9:17 pm, Jiri Gaisler wrote: > On 2/5/19 5:21 AM, Chris Johns wrote: >> On 4/2/19 9:49 pm, Jiri Gaisler wrote: >>> These two patches should be applied to gdb-8.2.1 to build sis for SPARC and >>> RISC-V: >> Thanks. >> >>> https://gaisler.se/gdb/gdb-8.2.1-riscv-config.patch >>> >>> https://gaisler.se/gdb/gdb-8.2.1-sis-2.11.patch >> I see in this patch `erc32` has been removed and replaced with `sis`. What >> does >> this mean? >> >> I have attached the patches to #3460. The SIS patch had to be compressed. > > erc32 is a bit misleading since sis also emulates leon2/3 and riscv.
OK. We made a move from the sis bsp to erc32 recently in the documentation as the base example bsp. > To keep the files together, I created the new directory sis. The old erc32 > directory is no longer maintained. I don't think it will be possible to merge > sis in the foreseeable future. The gdb maintainers prefer simulators that use > the built-in sim engine (sim/common) and discourage custom sims like sis. OK. > I will maintain sis out-of-tree for some time and then maybe make it > self-standing using the gdbserver interface. The libserver code in RTEMS is under a libsdb license if it is in anyway useful. > However, sis uses several binutils libraries (bfd, opcodes, readline) which > makes it convenient to build it in the gdb tree. Having the simulator > built-in to gdb is also convenient since we don't need to start a separate > process and connect to some (hopefully) free port.... QEMU works this way and I have not found the separation an issue. What I do like is the console not being messed up. When I run gdb psim or sis inside emacs, which is using MI, I do not get the console output in the IO window which is a pain. Chris _______________________________________________ devel mailing list devel@rtems.org http://lists.rtems.org/mailman/listinfo/devel
