Age | Commit message (Collapse) | Author | Files | Lines |
|
12 insertions(+), 10651 deletions(-)
I think that says it all :-)
|
|
When we have an enum that is used as a union discriminator, what
matters is that the scalar mappings are the same, not if the types are
the same (otherwise we get warnings about uint1632).
Thanks to gd for noticing this.
|
|
TDB doesn't have NDR64, but it needs to know how to map the new types
from pidl
|
|
The default enum in NDR63 is 32 bits, not 16 bits. We need a uint1632
type to get the alignment right.
|
|
|
|
List,
Please review this patch to pidl.
Basically, we need to process the wireshark conformance file BEFORE
we process the idl file since this file may define external types
and set the alignment for them (using the TYPE directive).
Otherwise pidl will default all external types to use 4byte alignment
which breaks (much more often) on NDR64
regards
ronnie sahlberg
From 8f86903fc353d0906bd82e72ce19c5af09beb001 Mon Sep 17 00:00:00 2001
From: Ronnie Sahlberg <ronniesahlberg@gmail.com>
Date: Mon, 5 Oct 2009 15:22:43 +1100
Subject: [PATCH] In the PIDL wireshark backend, we define external types in the conformance
file using the TYPE directive.
If we declare external types here, we must parse this file before we process
the IDL file, or else these external types will all default to 4byte padding
(pidl assumes all unknown types are 4byte aligned).
Make sure we read the conformance file and create these new types before
we parse the idl file.
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
|
|
This is the type used for a variable that is 32 bits for NDR32 and 64
bits for NDR64
|
|
NDR64 has a 'trailing gap' alignment, which aligns the end of a
structure on the overall structure alignment.
This explains the discrepancy we had with the RPC-SAMR test and NDR64
|
|
This fixes the problem with samr UserInfo16 when NDR64 is enabled
|
|
macro.
|
|
Caught by Guenther.
|
|
This allows for easier implementation of the NDR32/NDR64 split
|
|
Added support for NDR64 to the samba4 pidl generator
|
|
Jeremy.
|
|
to the C standard an enum is guarenteed to be an (int), which means
for 4 byte ints specifying a type of 0x80000000 is an invalid value.
The Solaris compiler complains about this. Fix by adding an (int)
cast in front of the value generation.
Jeremy.
|
|
|
|
metze
|
|
metze
|
|
metze
|
|
functions
metze
|
|
ParseFunction as glue
This prepares the following ParseFunctionAsync* commits.
metze
|
|
ParseOutputArgument()
metze
|
|
metze
|
|
This tries to fix commit 8eff9f9a3167eb0c2a4c00edf5a4cdbbc06c4dfd.
The problem was:
Undefined subroutine &Parse::Pidl::Samba4::Python::mapType called at /home/metze/master4/pidl/lib/Parse/Pidl/Samba4/Python.pm line 848.
Jelmer: please check this.
metze
|
|
This fixes some problems in the samr Python bindings that pidl was
(correctly) warning about.
|
|
Use py_talloc_reference in DCE/RPC code, fixes
access to SAMR pipe.
|
|
Original commit message in wireshark SVN (rev 28961):
====
From Kovarththanan Rajaratnam:
More "Cleanup header_field_info definitions"
====
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
|
|
metze
|
|
midl also supports this:
struct {
long l1;
[string] wchar_t str[16];
long l2;
};
Where the wire size of str is encoded like a length_is() header:
4-byte offset == 0;
4-byte array length;
The strings are zero terminated.
metze
|
|
'HASH(0x....)'.
|
|
than spewing pages of perl warnings.
|
|
Signed-off-by: Jelmer Vernooij <jelmer@samba.org>
|
|
It was creating new objects with bogus data.
This reverts commit 210830392999691469a55f4f5e2032abcdb20b8a.
|
|
For samba4 you need one of this:
samba/source4# make full_idl
samba/source4# make clean
samba# git clean -x -f
metze
|
|
metze
|
|
|
|
Code isn't generated to iterate over arrays of length 0, but the
variable declarations still are. The result is 'unused variable'
warnings. This only seems to be happening in one place right now, so I
targeted the fix to this case, but refactoring could be done to make
all variable declarations do this zero length check. Making it the
default would be a much more invasive fix.
Jelmer, please check!
|
|
(bug #6140)
|
|
|
|
|
|
|
|
This disgusting hack works around the fact that
ndr_pull_struct_blob_all will always fail on structures with relative
pointers. So, map ndr_unpack to ndr_pull_struct_blob_all only if we
don't have any relative pointers in this structure.
Andrew Bartlett
|
|
ndr_interface_table
metze
|
|
This breaks (some?) files with gcc.
This reverts commit 98a4327b19e83ffad4e0be7e8895fdbd9e48d49f.
|
|
When compiling Samba 4's IDL files buildsystem assumes that compiler
knows about -xc, which is a gcc spesefic switch as this switch isn't
needed on gcc either the flag was uneeded and can be removed. Thus no
warnings on non gcc systems.
Signed-off-by: Torgeir Lerkerod <Torgeir.Lerkerod@gmail.com>
Signed-off-by: Jelmer Vernooij <jelmer@samba.org>
|
|
bug in pidl where pidl uses "const" in a struct where there should not be
const.
There is only a single place in the Samba codebase where this triggers a
warning.
|
|
|
|
../librpc/gen_ndr/cli_spoolss.c: In function `rpccli_spoolss_EnumPrinterData':
../librpc/gen_ndr/cli_spoolss.c:3400: warning: passing arg 1 of `memcpy' discards qualifiers from pointer target type
This is a very similar fix to:
2337f402f0f2fd6c1fa077b1b35347905a38c715
|
|
Jelmer, Metze, please check.
Guenther
|
|
Michael
|