[Libre-soc-bugs] [Bug 238] POWER Compressed Formal Standard writeup

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=238

--- Comment #69 from Jacob Lifshay <programmerjake at gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #68)
> (In reply to Jacob Lifshay from comment #67)
> 
> > Except that BE instruction encoding is incompatible with the PowerPC64LE ABI
> > which requires that instruction encoding is set to LE on call/return/unwind.
> 
> tck tck... thinks... does the dual halfword memory swapping similarly
> interfere with ppc64le ABI? i would expect the answer to that to be yes.

No it does not interfere, because it is changing both the order that bytes are
conceptually read from memory as well as swapping the instruction encoding so
that the combination matches the original unmodified 32-bit LE encoding.

> the *instruction* encoding is required to be LE: what about data?

that too, that's the whole point of the ppc64le ABI: the data is in LE format
to match the de-facto industry standard (because of x86 and Arm being LE).

> remember we are expecting that all code set to known-good state (mode) on
> entry/exit (at the ABI points).

yup!

> i would reaaally prefer that we not have to define a new ABI, it means
> recompiling entire OS Distros.

I completely agree, hence why I'm pushing against just having instructions be
in BE mode.

-- 
You are receiving this mail because:
You are on the CC list for the bug.