[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RFC v2 04/13] target/s390x: remove tcg-stub.c
From: |
Claudio Fontana |
Subject: |
Re: [RFC v2 04/13] target/s390x: remove tcg-stub.c |
Date: |
Tue, 20 Apr 2021 15:00:33 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.12.0 |
On 4/20/21 2:54 PM, David Hildenbrand wrote:
> On 20.04.21 12:36, Claudio Fontana wrote:
>> now that we protect all calls to the tcg-specific functions
>> with if (tcg_enabled()), we do not need the TCG stub anymore.
>
> You need compile-time checks, not runtime checks. Any calls have to be
> protected by #ifdef, otherwise the compiler might bail out.
This is not true though, tcg_enabled() is #defined as 0 if tcg is not enabled.
#define kvm_enabled() (0)
Compiler will elide the code if after the preprocessor pass the code is:
if (0) {
}
It adds the benefit of actually checking the syntax of the code inside.
As long as the prototypes are in sight, we rely on this for i386 and ARM
already, to avoid accumulating stubs.
>
> Maybe you just wanted to state it differently in this patch description.
>
Thanks,
Claudio
[RFC v2 03/13] hw/s390x: tod: make explicit checks for accelerators when initializing, Claudio Fontana, 2021/04/20
[RFC v2 05/13] target/s390x: start moving TCG-only code to tcg/, Claudio Fontana, 2021/04/20
[RFC v2 09/13] target/s390x: use kvm_enabled() to wrap call to kvm_s390_get_hpage_1m, Claudio Fontana, 2021/04/20
[RFC v2 11/13] target/s390x: move kvm files into kvm/, Claudio Fontana, 2021/04/20