qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
* which CPUs should call aarch64_add_{sve,pauth,sme}_properties() ?
@ 2023-06-26 16:54 Peter Maydell
  2023-06-27  6:22 ` which CPUs should call aarch64_add_{sve, pauth, sme}_properties() ? Richard Henderson
  0 siblings, 1 reply; 2+ messages in thread
From: Peter Maydell @ 2023-06-26 16:54 UTC (permalink / raw)
  To: QEMU Developers; +Cc: Richard Henderson

At the moment:
 * aarch64_a64fx_initfn() calls aarch64_add_sve_properties()
 * aarch64_max_tcg_initfn() calls all of
   aarch64_add_pauth_properties(), aarch64_add_sve_properties(),
   aarch64_add_sme_properties()

and it happens that (ignoring 'max') we only have a64fx implementing SVE
and no CPUs implementing Pauth or SME yet.

Is the idea that all CPUs (both ones trying to match real hardware
and ones which are QEMU-specific like 'max') should call these
functions to set up the properties if they exist? Or is the idea
that 'max' is special and match-real-hardware CPUs don't need the
user-facing configurability?

If the former, maybe it would be better to put these in
arm_cpu_post_init() so they're automatically called if the CPU
specific initfn set the ID register value that indicates the
presence of pauth/sve/sme.

If the latter, A64FX is an odd-one-out for creating the SVE props.

I ask because Neoverse-V1 has both SVE and Pauth, and I'm not
sure if its init fn should be calling these functions.

thanks
-- PMM


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2023-06-27  6:22 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-06-26 16:54 which CPUs should call aarch64_add_{sve,pauth,sme}_properties() ? Peter Maydell
2023-06-27  6:22 ` which CPUs should call aarch64_add_{sve, pauth, sme}_properties() ? Richard Henderson

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).