[v4,0/8] limit passing around of cpu_user_regs

Message

Jan Beulich Feb. 20, 2024, 8:51 a.m. UTC

Unlike (synchronous) exception handlers, interrupt handlers don't normally
have a need to know the outer context's register state. Similarly, the vast
majority of key handlers has no need for such.

1: serial: fake IRQ-regs context in poll handlers
2: keyhandler: drop regs parameter from handle_keyregs()
3: serial: drop serial_rx_fn's regs parameter
4: PV-shim: drop pv_console_rx()'s regs parameter
5: serial: drop serial_[rt]x_interrupt()'s regs parameter
6: IRQ: drop regs parameter from handler functions
7: x86/APIC: drop regs parameter from direct vector handler functions
8: consolidate do_bug_frame() / bug_fn_t

I think it is now only the lack of an ack for the 1st patch which is keeping
this series from going in.

Jan

Comments

Jan Beulich Feb. 20, 2024, 8:58 a.m. UTC | #1

On 20.02.2024 09:51, Jan Beulich wrote:
> Unlike (synchronous) exception handlers, interrupt handlers don't normally
> have a need to know the outer context's register state. Similarly, the vast
> majority of key handlers has no need for such.
> 
> 1: serial: fake IRQ-regs context in poll handlers
> 2: keyhandler: drop regs parameter from handle_keyregs()
> 3: serial: drop serial_rx_fn's regs parameter
> 4: PV-shim: drop pv_console_rx()'s regs parameter
> 5: serial: drop serial_[rt]x_interrupt()'s regs parameter
> 6: IRQ: drop regs parameter from handler functions
> 7: x86/APIC: drop regs parameter from direct vector handler functions
> 8: consolidate do_bug_frame() / bug_fn_t
> 
> I think it is now only the lack of an ack for the 1st patch which is keeping
> this series from going in.

Actually while sending I noticed that further patch 5 still is in need
of an Arm ack.

Jan