On Thu, May 22, 2025 at 15:09:24 +0200, Michal Prívozník wrote:
On 5/22/25 15:00, Peter Krempa wrote:
> On Thu, May 22, 2025 at 14:25:14 +0200, Michal Privoznik via Devel wrote:
>> From: Michal Privoznik <mprivozn(a)redhat.com>
>>
>> Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
>> ---
>> tools/nss/libvirt_nss.c | 15 +++++----------
>> 1 file changed, 5 insertions(+), 10 deletions(-)
>
> IIRC the nss plugin was meant to be kept free of both the libvirt
> library and glib as it's actually linked into the address space of other
> processes.
>
>>
>> diff --git a/tools/nss/libvirt_nss.c b/tools/nss/libvirt_nss.c
>> index 25e2ec0642..8460b63228 100644
>> --- a/tools/nss/libvirt_nss.c
>> +++ b/tools/nss/libvirt_nss.c
>> @@ -134,7 +134,7 @@ findLease(const char *name,
>>
>> DEBUG("Dir: %s", leaseDir);
>> while ((entry = readdir(dir)) != NULL) {
>> - char *path;
>> + g_autofree char *path = NULL;
>
> So while we're okay using __attribute__(cleanup) I don't think we can
> use glib here.
>
Please see my previous patch to understand why this is NOT grabbing glib in.
I see. Well it's extremely confusing to name the macro the same as in
glib. As you can clearly see it makes it seem that glib is used.
I suggest you pick a different name.