Re: [PATCH] x86/irq: do not insert IRQ_MSI_EMU in emuirq mappings

2023-11-02 Thread Stefano Stabellini
On Thu, 2 Nov 2023, Jan Beulich wrote: > On 31.10.2023 17:33, Roger Pau Monné wrote: > > On Tue, Oct 31, 2023 at 03:30:37PM +0200, Xenia Ragiadakou wrote: > >> Do not use emuirq mappings for MSIs injected by emulated devices. > >> This kind of pirq shares the same emuirq value and is not remapped.

Re: [PATCH] x86/irq: do not insert IRQ_MSI_EMU in emuirq mappings

2023-11-02 Thread Jan Beulich
On 31.10.2023 17:33, Roger Pau Monné wrote: > On Tue, Oct 31, 2023 at 03:30:37PM +0200, Xenia Ragiadakou wrote: >> Do not use emuirq mappings for MSIs injected by emulated devices. >> This kind of pirq shares the same emuirq value and is not remapped. > > AFAICT adding the extra emuirq mappings is

Re: [PATCH] x86/irq: do not insert IRQ_MSI_EMU in emuirq mappings

2023-10-31 Thread Roger Pau Monné
On Tue, Oct 31, 2023 at 03:30:37PM +0200, Xenia Ragiadakou wrote: > Do not use emuirq mappings for MSIs injected by emulated devices. > This kind of pirq shares the same emuirq value and is not remapped. AFAICT adding the extra emuirq mappings is harmless, and just adds an extra layer of translati

[PATCH] x86/irq: do not insert IRQ_MSI_EMU in emuirq mappings

2023-10-31 Thread Xenia Ragiadakou
Do not use emuirq mappings for MSIs injected by emulated devices. This kind of pirq shares the same emuirq value and is not remapped. Fixes: 88fccdd11ca0 ('xen: event channel remapping for emulated MSIs') Signed-off-by: Xenia Ragiadakou --- Question: is there any strong reason why Linux HVM gues