Closed mdchitale closed 3 years ago
Merged. Thanks.
Regards
Kumar
From: mdchitale @.> Sent: Tuesday, August 31, 2021 9:15 AM To: riscv/riscv-platform-specs @.> Cc: Subscribed @.**> Subject:* [riscv/riscv-platform-specs] ISA requirements (#49)
This patch merges the ISA requirements as crafted by Greg and the existing architecture requirements in the spec.
You can view, comment on, or merge this pull request online at:
https://github.com/riscv/riscv-platform-specs/pull/49 Commit Summary
File Changes
Patch Links:
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/riscv/riscv-platform-specs/pull/49, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABLRGVS67LGNLF3UPE2FYPLT7T5YTANCNFSM5DEPDE4Q . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
For what it's worth, some of these topics are contentious and shouldn't be assumed to sail through without a fight. cc @gfavor
In particular, requirements like "hardware support for all misaligned accesses" are entirely inappropriate, for two reasons:
Ageed. That's why it is only mandated for server extension. The base allows the hardware not to implement misaligned access. In fact, the base also requires the M-mode to handle misaligned load/store traps.
On Tue, Aug 31, 2021 at 12:33 PM Andrew Waterman @.***> wrote:
In particular, requirements like "hardware support for all misaligned accesses" are entirely inappropriate, for two reasons:
- There is no functional problem with emulating them in an M-mode trap handler; it's a performance problem.
- The entire concept of "hardware support" is vacuous. Are we trying to say e.g. that QEMU is not a valid implementation of this platform?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/riscv/riscv-platform-specs/pull/49#issuecomment-909546138, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA6CNH2RAO3UI4CG3YUULC3T7UVCFANCNFSM5DEPDE4Q .
-- Regards, Atish
Nevertheless, it seems inappropriate to mandate HW implementation, for both reasons I mentioned. At minimum, it implies that non-HW implementations like QEMU cannot conform to this spec, which is presumably not the intent.
Even setting that aside, mandating support for features that are readily emulated seems questionable to me, because it conflates a performance requirement for a functional one. Surely this platform is not going to mandate a minimum clock rate, IPC, or core count, right?
This patch merges the ISA requirements as crafted by Greg and the existing architecture requirements in the spec.
Signed-off-by: Greg Favor gfavor@ventanamicro.com Signed-off-by: Mayuresh Chitale mchitale@ventanamicro.com