Skip to content

multipath-tools series part IV: identify paths by dev_t

[email protected] dm-devel 2 versions – In review

Get series with b4
b4 am [email protected]

Versions newest first · linked by author and title

  1. v2 Latest2 of 54 patches 2 repliesIn review
  2. v112 of 54 patches 6 repliesSuperseded

2 of 54 patches are in this archive. The cover letter announces 54.

Patches in v2 No applies recorded yet

#Subject and filesSizeRepliesApplied
44/54libmultipath: adopt_paths(): don't bail out on single path failurelibmultipath/structs_vec.c multipathd/main.c+13 -71 reply
46/54libmultipath: path_discover(): explain pathinfo flagslibmultipath/discovery.c+5 -01 reply

Reviews trailers per patch

All 2 patches have a Reviewed-by or Acked-by.

PatchBenjamin Marzinski
44/54R-b
46/54R-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v2 thread

Cover letter [PATCH v2 00/54] multipath-tools series part IV: identify paths by dev_t

From: Martin Wilck <[email protected]>

Hi Christophe, hi Ben,

This is part IV of a larger patch series for multipath-tools I've been preparing. It's based on the previously submitted part III.

The full series will also be available here:
https://github.com/mwilck/multipath-tools/tree/ups/submit-2008
This part is tagged "submit-200812-4".

The dominant idea of the patches in this subseries is to move away from the device node name as the main path identifier towards the dev_t aka major/minor number. This is motivated by the fact that dev_t is used by the kernel device mapper for referring to devices, both maps and paths. If searching a device by dev_t fails, it's basically certain not to exist. DM calls with major/minor number may even succeed if the respective devnode has not been created yet. dev_t also works better for devices that aren't fully initialized yet. Of course we still track the device name, but mostly for log messages and user-directed output.

In short, using dev_t as primary identifier makes our code more robust.

The subseries also contains some unrelated fixes for functions I worked on while making these changes.

Changes v1 -> v2, as suggested by Ben Marzinski:

* 44/54: "libmultipath: adopt_paths(): don't bail out on single path failure"
- fail if adding a specific path was requested and failed.
* 46/54: "libmultipath: path_discover(): always set DI_BLACKLIST"
- dropped, replaced by "libmultipath: path_discover(): explain pathinfo flags", which contains only comments

NOTE: In my v1 submission, I made a mistake when sending part V, so that *patch number 54 is present twice* in the full series.
I deliberately didn't correct that this time, to preserve numbering.

Martin Wilck (12):
libmultipath: adopt_paths(): use find_path_by_devt()
libmultipath: adopt_paths(): don't bail out on single path failure
libmultipath: path_discover(): use find_path_by_devt()
libmultipath: path_discover(): explain pathinfo flags
libmultipath: get_refwwid(): use find_path_by_devt()
libmultipath: get_refwwid(): call get_multipath_config() only once
libmultipath: get_refwwid(): get rid of "check" label
libmultipath: get_refwwid(): use symbolic return values
libmultipath: get_refwwid(): use switch statement
libmultipath: constify get_mpe_wwid()
multipath: call strchop() on command line argument
libmultipath: get_refwwid(): skip strchop(), and constify dev
parameter

 libmultipath/config.c      |   2 +-
 libmultipath/config.h      |   2 +-
 libmultipath/configure.c   | 200 +++++++++++++------------------------
 libmultipath/configure.h   |   2 +-
 libmultipath/discovery.c   |  37 ++++---
 libmultipath/structs_vec.c |  17 ++--
 multipath/main.c           |   2 +
 multipathd/main.c          |   3 +-
 8 files changed, 103 insertions(+), 162 deletions(-)