s3: Fix streams enumeration bug in OneFS implementation
Previously, we didn’t call SMB_VFS_OPEN_DIR from the streams module,
instead we called fdopendir(). As a result we failed to populate the
dir_state list in the readdirplus module. So when we tried to view the
stream data, we will always returned NULL.
To solve this I separated onefs_opendir() and the initialization of
the dir_state list. This is done by introducing a new utility function
“onefs_rdp_add_dir_state()”, which initializes the dir_state structure
and adds it to the dir_state list. This function is called from the
streams module before calling readdir().
todd stecher [Thu, 12 Feb 2009 08:11:38 +0000 (00:11 -0800)]
S3: Stop creating SMBD cores when failing to create a pipe.
This was uncovered when the MAX FD limit was hit, causing an instant core
and invoking error reporting. This fix causes SMBD to exit, but without
building a core.
todd stecher [Wed, 18 Feb 2009 00:16:35 +0000 (16:16 -0800)]
S3: Allow SMBD processes to survive in low memory condidtions
This commit adds a configure argument which allows for setting MADV_PROTECT
in the madvise() API. With this enabled the kernel won't kill SMBD when
it's running low on memory.
Andrew Tridgell [Wed, 18 Feb 2009 06:37:45 +0000 (17:37 +1100)]
Worked around a problem with select/poll/epoll and gnutls
Our packet layer relies on the event system reliably telling us when a
packet is available. When we are using a socket layer like TLS then
things get a bit trickier, as there may be bytes in the encryption
buffer which could be read even if there are no bytes at the socket
level. The GNUTLS library is supposed to prevent this happening by
always leaving some data at the socket level when there is data to be
processed in its buffers, but it seems that this is not always
reliable.
To work around this I have added a new packet option
packet_set_unreliable_select() which tells the packet layer to not
assume that the socket layer has a reliable select, and to instead
keep trying to read from the socket until it gets back no data. This
option is set for the ldap client and server when TLS is negotiated.
This seems to fix the problems with the ldaps tests.
Andrew Tridgell [Wed, 18 Feb 2009 03:46:57 +0000 (14:46 +1100)]
fixed some of the TLS problems
This fixes two things in the TLS support for Samba4. The first is to
use a somewhat more correct hostname instead of 'Samba' when
generating the test certificates. That allows TLS test clients (such
as gnutls-cli) to connect to Samba4 using auto-generated certificates.
The second fix is to add a call to gcry_control() to tell gcrypt to
use /dev/urandom instead of /dev/random (on systems that support
that). That means that test certificate generation is now very fast,
which was previously an impediment to putting the TLS tests on the
build farm.
Volker Lendecke [Sat, 14 Feb 2009 17:01:20 +0000 (18:01 +0100)]
Fix an invalid typecasting
entry->num_of_strings is a uint16_t. Casting it with
(int *)&entry->num_of_strings
is wrong, because it gives add_string_to_array the illusion that the object
"num" points to is an int, which it is not.
In case we are running on a machine where "int" is 32 or 64 bits long, what
happens with that cast? "add_string_to_array" interprets the byte field that
starts where "num_of_strings" starts as an int. Under very particular
circumstances this might work in a limited number of cases: When the byte order
of an int is such that the lower order bits of the int are stored first, the
subsequent bytes which do not belong to the uint16_t anymore happen to be 0 and
the result of the increment still fits into the first 2 bytes of that int, i.e.
the result is < 65536.
The correct solution to this problem is to use the implicit type conversion
that happens when an assignment is done.
BTW, this bug is found if you compile with -O3 -Wall, it shows up as a warning:
rpc_server/srv_eventlog_lib.c:574: warning: dereferencing type-punned pointer
will break strict-aliasing rules
Zach Loafman [Sat, 14 Feb 2009 18:20:33 +0000 (18:20 +0000)]
s3 auth: Add parameter that forces every user through an NSS lookup
When set to yes, "force username map" forces every user, even AD
users, through an NSS lookup. This allows the token to be overridden
with information from NSS in certain broken environments.
Jeremy Allison [Mon, 16 Feb 2009 02:12:20 +0000 (18:12 -0800)]
Attempt to fix bug #6099. According to Microsoft
Windows 7 looks at the negotiate_flags
returned in this structure *even if the
call fails with access denied ! So in order
to allow Win7 to connect to a Samba NT style
PDC we set the flags before we know if it's
an error or not.
Jeremy.
projects / thirdparty / samba.git / log
