actual wmem_allocator_t structure. This simplifies the internal API and
deduplicates a few alloc/free calls in the individual allocator implementations.
I'd originally made the allocators responsible for this on purpose with the
idea that they'd be able to optimize something clever based on the type of
allocator, but that's clearly more work and complexity than it's worth given
the small number of allocators we create/destroy.
svn path=/trunk/; revision=49512
Improve documentation for tshark's -z io,stat somewhat so that it's clear(er)
that filters for the statistics must be associated with the -z flag and not
applied via -Y.
svn path=/trunk/; revision=49422
recurring callbacks, I suspect most other potential uses will be once-only, so
make that possible, and improve the documentation on the remaining issues.
Also separate out the code into its own files and the testing into its own
test case.
svn path=/trunk/; revision=49209
the behaviour emem has for seasonal trees, which is that the master tree
structure is not actually seasonal - it is permanent. When the seasonal memory
pool is cleared, the root node pointer in all of these permanent trees is set
to NULL, and the pool takes care of actually freeing the nodes.
Wmem can now mimic this by allocating the tree header struct in epan_scope(),
allocating any node structs in file_scope(), and registering a callback on
file_scope() that NULLs the pointer in the epan_scope() header. Yes, this is
confusing, but it seemed simpler than adding manual callback registrations to
every single dissector that currently uses seasonal trees.
The callbacks may also be useful for other things that need cleanup (I'm
thinking resource handles stored in wmem memory that need to be fclosed or
what-have-you before they the handle is lost).
As indicated by the number of caveats in README.wmem, the implementation
probably needs a bit of work to make it safer/saner/more-useful. Thoughts
(or patches!) in this direction are more than welcome.
svn path=/trunk/; revision=49205
It is useful to see not only the minimal, maximum and average service time for
RPC procedures, but also the total time these took.
From me: add it to the man page.
svn path=/trunk/; revision=49144
- better tests
- fix a bug caught by the better tests
- implement append_c and append_unichar, with tests
Wmem string-buffers now have feature parity with their emem equivalents, so
remove them from the TODO list.
svn path=/trunk/; revision=49060
Changes of note:
- Removed the 'Copied from' notice, it's only relevant if they're *not* using
the skeleton code. Added a paragraph to README.developer instead.
- Exorcised all references to if (tree) and placed them in their own section
at the bottom as an optimization. Hopefully this will be less confusing.
svn path=/trunk/; revision=48861
much easier to edit/maintain that way as well as much easier to copy for a
new dissector. Explicitly don't set the SVN id tag since this is a template.
svn path=/trunk/; revision=48860
Add a 2-pass display-filter flag to tshark so that reassembly and other forward-
looking dissections can be used with filters.
It's a bit of a hack, but this entire area of 2-pass analysis etc. is a giant
pile of hacks to begin with and needs cleaning up. For now just having this
feature is a big enough win.
svn path=/trunk/; revision=48589
configurable via command line options and print specific drop numbers
at the end. This allows to tune the buffering when using multiple
threads.
This also fixes a logic arrow in enforcing the limits.
The patch for the enhanced counters is from Anders Broman.
svn path=/trunk/; revision=48223
(removed in r48218) which did nothing particularly useful. Also lets us remove
another debugging environment variable.
svn path=/trunk/; revision=48219
Since there's now a suffix rule for the .1 man pages, don't create explicit
rules for them. (Leave a comment explaining why the .html and .4 rules need
to stay.)
Simplify CLEANFILES: there's no need to list each man page, just remove them
all.
svn path=/trunk/; revision=48135
This patch adds a new public API, proto_tree_add_bitmask_len(), identical to
proto_tree_add_bitmask() but using a caller-supplied length rather than an
inferred one. The underlying proto_item_add_bitmask_tree() code is modified
to display only fields for which all defined bits are available, and to
ignore bits that have no corresponding defined field ("forward compatibility"
cases).
From me: minor edits, see the bug for more details.
svn path=/trunk/; revision=48049
a GSList. This permits it to implement the new realloc and free functions. Also
fill in an empty gc function, since there isn't much it can do as far as
garbage-collection goes.
svn path=/trunk/; revision=47169
variable (WIRESHARK_DEBUG_USE_SLICES) which turns off the slab allocator and uses
g_slices instead (which can themselves be turned off by setting
G_SLICE=always-malloc).
This makes debugging problems in slab-allocated memory easier to find
(hopefully including https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=8197 ).
Set WIRESHARK_DEBUG_USE_SLICES when running Valgrind on *shark.
Remove unused structure member: emem_chunk_t.org.
svn path=/trunk/; revision=47110
shown for the previous input format (showing "data" as a pointer is a
bit confusing, as it's just in-line data in the input), and give a more
detailed description of the issues with that file format (the size of
the seconds field of the time stamp is platform-dependent).
svn path=/trunk/; revision=46864
an I or O indicating the direction of the packet. This
will be stored in the generated file if the pcap-ng
format is used.
Thanks to Jaap for suggesting to integrate this into
the preamble!
svn path=/trunk/; revision=46744
Leave pcap the default file format for now. The is should be reconsidered
before the next release as it might make sense to use pcapgn as
the default as we do for dumpcap. (We can use the -P option to allow
switching back to pcap).
svn path=/trunk/; revision=46691
use it as example in a few places and point out that if you're not using the
return value to build a subtree, you probably shouldn't be using the function.
svn path=/trunk/; revision=46617
determine the desired type. This has two advantages over the old way:
- just one environment variable for valgrind to override in order to guarantee
that ALL allocators use memory it can track, and just one place to check that
variable
- allocator owners no longer have to include headers specific to their
allocator, allowing them to change allocators without adjusting all their
#includes
svn path=/trunk/; revision=46604
Combine the two comments in the Portability section (which largely said the
same thing) on the perils of that function.
Don't suggest it as an option to ensure there are enough bytes in the TVB.
svn path=/trunk/; revision=46590
Add that option to tshark, too, and document it.
The option can't be given to Wireshark because the GUI already has a "-g"
(goto packet).
svn path=/trunk/; revision=46513
