summaryrefslogtreecommitdiff
path: root/hurd
diff options
context:
space:
mode:
Diffstat (limited to 'hurd')
-rw-r--r--hurd/translator/tmpfs/discussion.mdwn52
1 files changed, 52 insertions, 0 deletions
diff --git a/hurd/translator/tmpfs/discussion.mdwn b/hurd/translator/tmpfs/discussion.mdwn
index 1d441c7d..bdee0f78 100644
--- a/hurd/translator/tmpfs/discussion.mdwn
+++ b/hurd/translator/tmpfs/discussion.mdwn
@@ -376,3 +376,55 @@ License|/fdl]]."]]"""]]
<mcsim> Anyway all these are just translators, so there shouldn't be any
problems.
<mcsim> pinotree: works
+
+
+## IRC, freenode, #hurd, 2012-03-22
+
+ <mcsim> Hello. Is it normal that when i try to run du at directory where
+ translator is mounted it says that directory is 'Not a directory'? Here
+ are some examples with different filesystems: paste.debian.net/160699
+ First is ramdisk+ext2fs, second is tmpfs, third is ext2fs.
+ <civodul> i can't reproduce the problem with ext2fs
+ <civodul> perhaps you can try rpctracing it to see where ENOTDIR comes from
+ <mcsim> civodul: when I run du io_stat_request ipc is called. But reply is
+ ((os/kern) invalid address). Where is server code for this ipc? I only
+ found its definition in defs file and that's all.
+ <civodul> mcsim: server code is in libdiskfs + ext2fs, for instance
+ <mcsim> civodul: Does io_stat_request have changed name in server code? I
+ just can't find it. Here are my grep results fore io_stat_request (i was
+ grepping in root of hurd repository: paste.debian.net/160708
+ <youpi> remove _request
+ <youpi> it's just io_stat
+ <mcsim> youpi: thank you
+
+
+## IRC, freenode, #hurd, 2012-04-08
+
+ <mcsim> youpi: I've corrected everything you said, and pushed code to new
+ branch mplaneta/tmpfs/master-v2
+ <youpi> mcsim: all applied, thanks !
+ <youpi> I'll probably test it a bit and upload a new version of hurd
+ <youpi> mcsim: it seems to be working fine indeed!
+ <mcsim> youpi: thank you for all your reviews, suggestions you gave and
+ corrections you made :)
+ <youpi> and it seems translators indeed work there too
+ <youpi> hopefully it'll work to run the debian installer
+ <youpi> that'd permit to solve memory consumption
+ <pinotree> (so tmpfs works really fine now? great!)
+ <youpi> I could reboot with tmpfs on /tmp and build a package there, yes
+ <mcsim> youpi: yes, I've compiled several packages already, but it does not
+ give big advantage in performance.
+ <youpi> I wasn't really looking for performance, but for correctness :)
+ <youpi> are you using writeback for your /, actually ?
+ <youpi> argl, /run gets triggered before mach-defpager is started
+ <youpi> the X11 socket works there too
+ <youpi> gnu_srs: might your mouse issue with Xorg be related with vnc usage
+ too?
+ <youpi> it seems ENOSPC works fine too
+ <mcsim> youpi: as to writeback. I think yes, because default pager is asked
+ to write data only when this data is evicted.
+ <youpi> I'm talking about kvm
+ <mcsim> youpi: I use real computer.
+ <youpi> ok
+ <youpi> but that indeed means writeback of ext2fs works, which is a good
+ sign :)