[Freeipa-devel] [PATCHES 134-136] extdom: handle ERANGE return code for getXXYYY_r()

[Alexander Bokovoy]

On Thu, 05 Mar 2015, Sumit Bose wrote:
>On Thu, Mar 05, 2015 at 09:16:36AM +0100, Sumit Bose wrote:
>> On Wed, Mar 04, 2015 at 06:14:53PM +0100, Sumit Bose wrote:
>> > On Wed, Mar 04, 2015 at 04:17:55PM +0200, Alexander Bokovoy wrote:
>> > > On Mon, 02 Mar 2015, Sumit Bose wrote:
>> > > >diff --git a/daemons/ipa-slapi-plugins/ipa-extdom-extop/ipa_extdom_common.c b/daemons/ipa-slapi-plugins/ipa-extdom-extop/ipa_extdom_common.c
>> > > >index 20fdd62b20f28f5384cf83b8be5819f721c6c3db..84aeb28066f25f05a89d0c2d42e8b060e2399501 100644
>> > > >--- a/daemons/ipa-slapi-plugins/ipa-extdom-extop/ipa_extdom_common.c
>> > > >+++ b/daemons/ipa-slapi-plugins/ipa-extdom-extop/ipa_extdom_common.c
>> > > >@@ -49,6 +49,220 @@
>> > > >
>> > > >#define MAX(a,b) (((a)>(b))?(a):(b))
>> > > >#define SSSD_DOMAIN_SEPARATOR '@'
>> > > >+#define MAX_BUF (1024*1024*1024)
>> > > >+
>> > > >+
>> > > >+
>> > > >+static int get_buffer(size_t *_buf_len, char **_buf)
>> > > >+{
>> > > >+    long pw_max;
>> > > >+    long gr_max;
>> > > >+    size_t buf_len;
>> > > >+    char *buf;
>> > > >+
>> > > >+    pw_max = sysconf(_SC_GETPW_R_SIZE_MAX);
>> > > >+    gr_max = sysconf(_SC_GETGR_R_SIZE_MAX);
>> > > >+
>> > > >+    if (pw_max == -1 && gr_max == -1) {
>> > > >+        buf_len = 16384;
>> > > >+    } else {
>> > > >+        buf_len = MAX(pw_max, gr_max);
>> > > >+    }
>> > > Here you'd get buf_len equal to 1024 by default on Linux which is too
>> > > low for our use case. I think it would be beneficial to add one more
>> > > MAX(buf_len, 16384):
>> > > -    if (pw_max == -1 && gr_max == -1) {
>> > > -        buf_len = 16384;
>> > > -    } else {
>> > > -        buf_len = MAX(pw_max, gr_max);
>> > > -    }
>> > > +    buf_len = MAX(16384, MAX(pw_max, gr_max));
>> > >
>> > > with MAX(MAX(),..) you also get rid of if() statement as resulting
>> > > rvalue would be guaranteed to be positive.
>> >
>> > done
>> >
>> > >
>> > > The rest is going along the common lines but would it be better to
>> > > allocate memory once per LDAP client request rather than always ask for
>> > > it per each NSS call? You can guarantee a sequential use of the buffer
>> > > within the LDAP client request processing so there is no problem with
>> > > locks but having this memory re-allocated on subsequent
>> > > getpwnam()/getpwuid()/... calls within the same request processing seems
>> > > suboptimal to me.
>> >
>> > ok, makes sense, I moved get_buffer() back to the callers.
>> >
>> > New version attached.
>>
>> Please ignore this patch, I will send a revised version soon.
>
>Please find attached a revised version which properly reports missing
>objects and out-of-memory cases and makes sure buf and buf_len are in
>sync.
ACK to patches 0135 and 0136. This concludes the review, thanks!

-- 
/ Alexander Bokovoy