On 02.08.2017 21:10, Maxime Villard wrote: > Le 02/08/2017 à 16:42, Kamil Rytarowski a écrit : >> On 02.08.2017 15:27, Johnny Billquist wrote: >>> On 2017-08-02 14:44, Maxime Villard wrote: >>>> Le 02/08/2017 à 14:21, Kamil Rytarowski a écrit : >>>>> On 02.08.2017 11:34, Maxime Villard wrote: >>>>>> Hi, >>>>>> I see that the Vax GENERIC kernel has COMPAT_IBCS2 enabled by >>>>>> default. >>>>>> But what >>>>>> exactly is it doing here? iBCS2 is a standard designed for x86 CPUs, >>>>>> and >>>>>> it has >>>>>> nothing to do with Vax. >>>>>> >>>>>> Is there something that I'm missing? If so, it would be good to >>>>>> enlighten the >>>>>> man page. >>>>>> >>>>>> Thanks, >>>>>> Maxime >>>>> >>>>> This looks like SVR3 compatibility, x86 variation of it is called >>>>> IBCS2. >>>> >>>> Yes, I guessed that when reading the entry in GENERIC; but shouldn't >>>> it be >>>> renamed, or at least, made a little less confusing? In the man page for >>>> example, we could say that iBCS2 stands for SVR3 on Vax, even if it is >>>> called >>>> iBCS2... >>> >>> I would vote for renaming. iBCS2 is something that is totally cryptic to >>> me. SVR3 is something I would expect most people to understand what >>> it is. >>> But I suspect that also means renaming it for the x86 folks, and I >>> wonder how easy it might be to get them onboard. >>> >>> Johnny >>> >> >> I would keep it as it is, and improve documentation. >> > > That's what I would do too. > > In fact, I don't understand for which specific purpose compat_ibcs2 was > written. Was it written to be a compat for SVR3, but called iBCS2 for some > reason? (in which case, it would be suspicious to have it on i386.) Or > was it > intended to be the real x86-specific iBCS2? (in which case, it would be > suspicious to have it on Vax.) > It started as compat_ibcs2 which was SVR4 (ELF) i386. It was extended with SCO UNIX SVR3 (IBCS2 COFF) i386. Later there was added SCO UNIX SVR3 (IBCS2 XAOUT) i386. The compat_ibcs2 i386 code was reused for DEC VAX SVR3. > The man page is not particularly enlightening, and I fear that in order to > figure out which compat (or which mix of both) compat_ibcs2 was meant to > be, > we will have to read the code and compare it with the iBCS2 > specification. I > couldn't find this spec on the internet. > It started as SVR4 i386. IBCS2 ELF was renamed to compat_svr4, and compat_ibcs2 was used for i386 SVR3. I think that the ibcs2 evaluated as an alias for SVR3. I don't know whether somebody reused this name on original VAX UNIX. I proposed to keep it as it is, because there are more aliases like i386 that have not clear meaning. > Maxime
Attachment:
signature.asc
Description: OpenPGP digital signature