On 17.12.2019 17:33, Andrew Cooper wrote: > On 17/12/2019 16:17, Jan Beulich wrote: >> On 13.12.2019 20:04, Andrew Cooper wrote: >>> Andrew Cooper (6): >>> x86/suspend: Clarify and improve the behaviour of do_suspend_lowlevel() >>> x86/suspend: Don't bother saving %cr3, %ss or flags >>> x86/suspend: Don't save unnecessary GPRs >>> x86/suspend: Restore cr4 later during resume >>> x86/suspend: Expand macros in wakeup_prot.S >>> x86/suspend: Drop save_rest_processor_state() completely >>> >>> xen/arch/x86/acpi/suspend.c | 55 ++-------------- >>> xen/arch/x86/acpi/wakeup_prot.S | 136 >>> ++++++++++++---------------------------- >>> xen/arch/x86/boot/wakeup.S | 2 +- >>> 3 files changed, 46 insertions(+), 147 deletions(-) >> Based on Roger's review >> Acked-by: Jan Beulich <[email protected]> >> >> One remark on the combination of patches 2 and 5: The loading of >> the stack related registers would now seem to be a fair candidate >> for using LSS (generally to be preferred over MOV-to-SS). > > Well... You've just fixed c/s ffa21ea5303 in the emulator, and it > demonstrates why LSS can't be used.
Hmm, indeed, how did I forget? (It's really very counter-intuitive for this insn to not be universally usable.) Jan _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
