02:13:14 [illumos-gate] 17223 move snoop parsing out of mac_ktest -- Patrick Mooney 13:13:52 If I do an open O_RDONLY, is there any way to later change the file descriptor you get to O_RDWR? 13:15:32 As far as I can tell, from reading the manual and testing it, fcntl(F_SETFL) doesn't work for this (although it might do on HPUX, of all things). 13:25:15 #define FMASK 0xa0ff /* all flags that can be changed by F_SETFL */ 13:25:37 by that mask, you should be able to add FWRITE 13:27:24 the libc has code to add FAPPEND. but I guess, it may also depend on file system... 13:53:24 Doesn't work, which is what the manual and the headers are explicit about. I was looking for an alternative route. 14:08:58 ptribble: no - that would break the security model around file descriptors-as-capabilities. 15:28:22 danmcd: Is it a known problem that creating NFSv4.1 XATTR files fails with NFS4ERR_PERM ? 15:29:06 Yeah, Cedric B filed some bugs, go search for NFS4ERR_PERM in illumos.org/issues 15:29:13 https://illumos.org/issues/ 15:30:29 Also... I thought XATTR was part of 4.0, not 4.1. I've been seeing some recent MacOS vs. NFS crap related to that, but only from my backup program. I should probably figure out how to do proper modern SMB (SMB3?) configurations on my backup volumes. 15:32:19 NFSv4.1 XATTRS are part of NFSv4.1 and NFSv4.2. Just some unhappy Linux people are pretty agressive about getting rid of it, usually by declaring anything which doesn't fit their policy as "deprecated" 15:33:00 FreeBSD is working on NFSv4.1/v4.2 XATTR support, and AFAIK the dcache NFS server supports them too 15:34:27 OpenText's NFSv4 driver for Windows supports "Alternate Data Streams" on Windows via NFSv4 XATTRs, and put Win32 EAs (Extended Attributes) into NFSv4 XATTRs too 15:34:34 So yes, there are users for this stuff 15:36:03 danmcd: Setting, getting, deleting NFSv4.1 attributes with ms-nfs41-client works with Solaris 11.4 NFSv4.1 nfsd, so this seems to be an Illumos issue 16:24:02 danmcd: "I thought XATTR was part of 4.0, not 4.1" <--- does that mean there is code to disable the NFS XATTRs ? 16:27:39 You can set your NFS server max/min versions. 16:28:17 Yes, I know, I tried both NFSv4.1 and NFSv4.2, but the ms-nfs41-client does not support NFSv4.0 17:53:53 I can never remember the gold-standard `wsdiff` invocation between two proto areas. 17:56:52 Does anyone here? 18:15:50 I use `wsdiff -r -uvV` - I don't know if that's the gold standard. 18:16:02 I just let nightly invoke it 18:17:02 anyone here claim to understand fmtopo-related things? context: https://code.illumos.org/c/illumos-gate/+/4107 ; specifically the "fabric-xlate" module. 18:17:02 → CODE REVIEW 4107: 17213 fmd crashes with NULL fabric nodeTab (NEW) | https://www.illumos.org/issues/17213 18:19:53 it's crashing for at least two users and it looks like fab_update_topo is missing an fflush - so it will feed itself broken xml unless the xml happens to be a multiple of the stdio buffer size in length.. 18:21:00 but I haven't managed to figure out how to get fmd to call that module's recv entry point. 18:26:03 andyf: works for me. (I'm comparing two should-be-VERY-similar builds of SmartOS, changing our required pkgsrc revision soon.) 18:56:39 danmcd: wsdiff -uVvr foo.log 18:57:02 sommerfeld: I don't, but I think rmustacc might 18:57:29 Assuming those flags aren't positional, you and Andy both are in agreement. Once my builds finish I'll be doing just this. Thanks! 19:01:15 Jordan Hendricks (who's at Oxide) mucked around in one of the two files in question @sommerfeld, but not in the function you're fixing. Dunno if any of that is in her L1/L2 cache. 19:02:01 Oh, looks like #omnios has the rest of this @sommerfeld story. :) 19:52:38 I'm looking at the omnios-extra repository on Github, specifically the top level README.md which covers using the build system. There are some pieces which I'm having trouble in understanding, despite looking through the files under /lib. Is there any place where I might find something more descriptive about using the build system?