| Subject: | Re: cloning devices and DTYPE_MISC |
|---|---|
| From: | christos@xxxxxxxxxx (Christos Zoulas) |
| Date: | Tue, 22 Jul 2008 19:19:26 UTC |
| Newsgroups: | fa.netbsd.tech.kern |
In article <200807221501.m6MF1eXr014213@xxxxxxxxxxxxx>, Matthew Mondor <mm_lists@xxxxxxxxxxxxxxx> wrote: >On Monday, 21 Jul 2008 5:43:16 >Thor Lancelot Simon <tls@xxxxxxxxxxxxxxx> wrote: > >> fd_clone() explicitly sets the type of the cloned file to >> DTYPE_MISC.One annoying effect this has is that it makes it hard >> to look at a processthat has several cloned devices open and see >> which is which. > >> It also makes it very hard for pseudodevices which have ioctls >> thatoperate on file descriptors to tell which are "their own" >> though thereare other means of dealing with that. > >Speaking of fdclone(), I have the pending kern/37878 PR which also >relates to DTYPE_MISC; > >Now that I quickly checked fstat(1) again I noticed that there are >cases for various other DTYPE_ although not for DTYPE_MISC, and that >it skips unknown types (only printing a warning about the descriptor >if -v switch is used). > >I'm not sure if fstat(1) or fdclone(2) should be corrected about this >or both (procfs does list those fine). I fixed that at 2008-04-04 although the fix is sub-optimal. I.e. no information is being told about DTYPE_MISC. christos |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: cloning devices and DTYPE_MISC, Matthew Mondor |
|---|---|
| Next by Date: | Re: cloning devices and DTYPE_MISC, Jason Thorpe |
| Previous by Thread: | Re: cloning devices and DTYPE_MISC, Matthew Mondor |
| Next by Thread: | Re: cloning devices and DTYPE_MISC, Jason Thorpe |
| Indexes: | [Date] [Thread] [Top] [All Lists] |