This is the mail archive of the libc-alpha@sources.redhat.com mailing list for the glibc project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
On Tue, May 27, Ulrich Drepper wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Thorsten Kukuk wrote: > > > With this, everything should work. > > The patch doesn't look bad but is the functionality really equivalent? Yes, it is. I only removed duplicated code from nss_compat (the same code was already in nss_nis or nss_nisplus). And all my tests shows the same result and I'm using it already here. > I'm worried about now using the NIS NSS module from cmopat instead of > going directly. Are there any ways there are differences? I wouldn't > expect since I hope the code is thread-safe. There are two problems: The NIS interface between the client and the server is not thread safe. As result, it could always be that the data is incomplete and users are missing or duplicated. But this is nothing new, this happens already with the old nss_compat module and it also happens with nss_nis. But the second problem I worry much more about: The new nss_compat module breaks RPM. What happens is: RPM is linked static and calls NSS functions. rpm dlopen nss_compat. nss_compat dlopen nss_nis. The second dlopen in nss_compat seg.faults, if the main program is linked static (I will write a short test case for this later). So, since we don't support dlopen in static linked programs official, is the seg.fault expected to be happen or is this a bug somewhere? If this is the expected behavior and not fixable, I think we should not use the new nss_compat code. I doubt that RPM will be fixed to not call NSS functions and be linked static. > What is in any case missing is cleanup. Breaking long lines, correct > line breaks, ... If you do this I'll check the changes in any we can > see how they perform. There is a lot of more cleanup missing and the initgroups_dyn() support is missing. Without this, login will be really slow if you have big NIS, NIS+ maps or whatever else. I will do this, after the above static linking/dlopen problem is resolved. Thorsten -- Thorsten Kukuk http://www.suse.de/~kukuk/ kukuk@suse.de SuSE Linux AG Deutschherrnstr. 15-19 D-90429 Nuernberg -------------------------------------------------------------------- Key fingerprint = A368 676B 5E1B 3E46 CFCE 2D97 F8FD 4E23 56C6 FB4B
Attachment:
pgp00000.pgp
Description: PGP signature
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |