On 1/19/26 18:50, Michael Tokarev wrote:
> On 1/19/26 04:28, Richard Henderson wrote:
>> On 1/14/26 00:00, Michael Tokarev wrote:
>>> Since CONFIG_EPOLL is now unused, it's okay to
>>> perform this rename, to make it less ugly.
>>>
>>> Since epoll is linux-specific and is always
>>> present, we might as well make CONFIG_EPOLL equal
>>> to CONFIG_LINUX.
>>>
>>> Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
>>> ---
>>> meson.build | 2 +-
>>> util/aio-posix.h | 4 ++--
>>> util/meson.build | 2 +-
>>> 3 files changed, 4 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/meson.build b/meson.build
>>> index f8b4f06049..552b34f34e 100644
>>> --- a/meson.build
>>> +++ b/meson.build
>>> @@ -2713,7 +2713,7 @@ config_host_data.set('CONFIG_INOTIFY1', have_inotify_init1)
>>> # has_header_symbol
>>> config_host_data.set('CONFIG_BLKZONED',
>>> cc.has_header_symbol('linux/blkzoned.h', 'BLKOPENZONE'))
>>> -config_host_data.set('CONFIG_EPOLL_CREATE1',
>>> +config_host_data.set('CONFIG_EPOLL',
>>> cc.has_header_symbol('sys/epoll.h', 'epoll_create1'))
>>
>> This is present in glibc 2.9, which is required.
>> Therefore you can drop this bit of configuration as well.
>
> It's a bit more tricky, see my comment in the patch description
> above. Yes, epoll_create1 is now always present *on linux*.
> But this particular test (cc.has_header_symbol) is run on all
> platforms, not only on linux. So if we drop this test, we'll
> have to alias CONFIG_EPOLL to CONFIG_LINUX. Because epoll is
> used outside of linux-user too (eg in the main loop), and there,
> the test is CONFIG_EPOLL, not CONFIG_LINUX.
>
> So it might be possible to replace this test with something
> like
>
> config_host_data.set('CONFIG_EPOLL', config_host_data.get('CONFIG_LINUX')
>
> but not drop it entirely.
From the patch description, I had expected s/CONFIG_EPOLL/CONFIG_LINUX/.
r~