Alex Rousskov [Mon, 26 Nov 2012 18:54:43 +0000 (11:54 -0700)]
Make it possible to match empty header field values using req_header and rep_header.
Warning: Some req_header and rep_header ACLs that were [accidentally] not
matching empty headers (e.g., "^$" or ".*") will now start matching them.
A new HttpHeader::getByNameIfPresent() method is added to be able to detect
presence of empty header fields while ACLHTTPHeaderData::match() is adjusted
to convert undefined String values into empty c-strings ("") for
ACLRegexData::match() to work.
Prior to these changes, when trying to match an empty header value with a
regex like "^$", ACLHTTPHeaderData::match() would return false because:
* HttpHeader::getStrOrList() and getByName() return an undefined String.
* String::termedBuf() returns NULL for undefined Strings; and
* ACLRegexData::match() always fails on NULL c-strings.
Amos Jeffries [Sun, 18 Nov 2012 10:37:19 +0000 (03:37 -0700)]
Fix several buffer termination bugs
* strcpy() replaced in several places with strncpy() to ensure destination
buffers are not overflowed.
* strncpy() does not nul-terminate the destination when the string being
copied in exactly fills the buffer. Ensure we have terminated strings
where it may matter.
Detected by Coverity Scan. Issues 740309, 740310, 740311, 740481, 740483
Amos Jeffries [Sun, 18 Nov 2012 10:02:38 +0000 (03:02 -0700)]
Remove MemPoolChunked::memPID
This member variable appears to have been missed when MemPool was split
into generic framework and specific Chunked implementation.
(rev:10513.1.1 aka trunnk rev:10517)
The memPID and its maintenance code was moved into MemImplementingAllocator
but this definitio left here un-initialized and shadowing the framework
member.
Amos Jeffries [Sun, 18 Nov 2012 07:06:26 +0000 (00:06 -0700)]
negotiate_kerberos_auth: better bounds checking
* sysconf() may return -N values on some platforms or values larger than
the hard-coded 1024 buffer size for hostname. Use sizeof() instead
since the buffer is hardcoded anyway.
* also, use return instead of exit() on the test binary to reduce
warnings from static analysis compilers.
Amos Jeffries [Sat, 17 Nov 2012 12:58:59 +0000 (05:58 -0700)]
Various memory leaks in configuration parsing
This lot are all small issues derived from allocating new memory and
assigning to a pointer already pointing at previous allocation, or
passing xstrdup() output to a caller which does not directly hold the
passed memory.
Both cases will disappear once we clean up teh string handlign in Squid
but for now these still need fixing to avoid leaking memory.
Detected by Coverity Scan. Issues 740430, 740432, 740439.
Amos Jeffries [Fri, 16 Nov 2012 06:45:56 +0000 (23:45 -0700)]
Fix various assertion with side effects
When compiled with high optimization and assert disabled these operations
would have disappeared. The side effects being:
* Disk I/O failure protection disabled. Allowing loops in diskd write.
* squidpurge error handling on command line parse gone. Causing segfault.
* squidpurge 'I am Alive' ticker feature cease working.
Detected by Coverity Scan. Issues 740299, 740300, 740301, 740302, 740303
Amos Jeffries [Thu, 15 Nov 2012 09:00:29 +0000 (02:00 -0700)]
ntlm_fake_auth: Fix nesting error
Broken macro wrapping leads to the fake authenticator sending bad
responses to Squid. This can ead to users being rejected by the fake
helper whose purpose is to accept everything.
Amos Jeffries [Wed, 14 Nov 2012 01:33:49 +0000 (18:33 -0700)]
digest_edirectory_auth: improved error handling
Malicious response from LDAP server can cause squid helper to crash.
Missing realm value returned from LDAP without error/missing value being
indicated in the response can lead to strcmp() using a NULL pointer.
Extremely unlikely to happen in practice, but worth fixing.
Changed HERE macro to be a true nop, clarified documentation, and marked is as deprecated.
Fixed syntax bug in comm.cc unearthed by the macro-to-function change.
- Try to update the index file in all cases the database modified
- The find operator in database should not modify the database. Currently
if an entry is expired, ssl_crtd removes the cert file but does not
update the index file.
rows. Currently we are using the new operator.
- Fix a small memory leak when remove entries from database: A row object
removed from TXT_DB indexes but never released. This patch:
* Use OPENSSL_malloc and OPENSSL_free to allocate/release memory for
TXT_DB rows. OpenSSL SDK assumes that always allocated using these
functions.
* Add code in Ssl::CertificateDb::Row destructor to correctly release
a TXT_DB row.
* Add the sq_TXT_DB_delete and sq_TXT_DB_delete_row functions which
removes a row from TXT_DB indexes.
With the patch r12422 which added support for OK/ERR/BH helper response codes,
the ssl_crtd interface become broken. The ConnStateData::sslCrtdHandleReply
ssl_crtd reply handler, expects a response code included at the beggining of
the returned message. This is not true after the changes.
This patch add a parameter in Ssl::CrtdMessage constructor to define a message
as "ssl_crtd reply" or "ssl_crtd request" message. For ssl_crtd reply messages
the message parser does not expect helper response code.
Amos Jeffries [Tue, 30 Oct 2012 09:16:33 +0000 (22:16 +1300)]
Fix HelperReply(const HelperReply&) is private
GCC 4.1 and possibly others use copy-constructor when an instance is
created in a function parameter list using copy by reference.
We must create such objects as loal variables and pass to the function
for best efficiency.
Amos Jeffries [Mon, 29 Oct 2012 23:12:04 +0000 (12:12 +1300)]
Support OK/ERR/BH response codes from any helper
Updates the helper reponse callback API from using char* buffer to a
HelperReply object storing teh response code, a blob buffer, and
pointer to the responding helper 'server' (if stateful).
* the helper I/O read handler is updated to parse the result code off
the start of the helper response as is currently done for channel-ID.
The callback handlers are altered to use the HelperReply::status instead
of parsing it off themselves individually.
* the remaining I/O read buffer is stored in a MemBuf and callbacks are
updated to use it via the method other().
* the responding helper-server is stored into the HelperReply object and
stateful helper callbacks are combined into the same API as stateless.
The callback handlers are updated to use HelperReply::lastserver instead
of function parameter.
After this patch the helper response format is:
[channel-ID] SP [result] [ [SP] blob] <terminator>
'SP' being one octet \0x20 character.
The behavour changes expected from this is that all helpers are now able
to send OK/ERR/BH states. Although the handlers for some helpers will
deal with the new states as unknown response. None of the bundled
helpers have yet been altered to make use of this changed potential.
TODO:
* implement key=value parser for the blob area of the format, and update
handlers to use the HelperReply API to retrieve them.
* upgrade helpers to make use of new response format
Amos Jeffries [Mon, 29 Oct 2012 04:59:58 +0000 (22:59 -0600)]
Polish: update the RefCount API a bt and split Lock out
* Shuffle RefCount.h and its unit-tests into src/base/
* Reworks struct Refcountable_ into class LockableObject in its own header
+ changing the reference counter accessors to a lock()/unlock() names
+ some minor symbol updates of code directly utilizing the RefCountable_
members
With this we can begin the process of replacing our multiple different
implementations of the reference-counting pattern using LockableObject.
No code changes have been made. Just symbol polishing.
TODO: update the unit-tests for refcounting to use CPPUnit
This patch adds note option to squid.conf. It is similar to
adaptation_meta but is applied after all adaptation and before logging.
Values of key/value pair notes can be logged using %{key}note macros.
note key value acl ...
logformat myFormat ... %{key}note ...
This option may be initially used to log custom information about the master
transaction. For example, an admin may configure Squid to log which "user group"
the transaction belongs to, where "user group" will be determined based on a
set of ACLs and not [just] authentication information.
From user point of view, adaptation_header sets/implies meta (i.e.,
setting adaptation_meta is sufficient to be able to log it using
%note) but the note option itself (if any) is evaluated later,
so it has no effect on ICAP headers.
Amos Jeffries [Fri, 26 Oct 2012 11:36:45 +0000 (05:36 -0600)]
Polish: Http::MethodType upgrade
Convert the METHOD_* enum into a class with accessors to determine
method string/image, idempotency, safety (as per RFC 2616 terminology).
* creates the initial http/ directory and http/libsquid-http.la
library as per SourceLayout needs
- libsquid-http is chosen to prevent symbols mistakes with libhttp
which may exist on some systems under the libhttp or http/libhttp name.
enum changes:
* migrates _method_t to Http::MethodType and makes it a enum with
auto-generated strings array provided by the new library.
* the enum list is extended to contain all methods in the HTTPbis WG
new method registry
- documentation now references that registry and HTTPbis draft
document listing methods
HttpRequest class changes:
* cacheable() changed to maybeCacheable() accessor
- to reflect that it determines NON-cacheability with certainty, but
when returns TRUE it can still be overridden by Http Reply details
and/or local environment details.
- updated to cache the complex calculation result.
HttpRequestMethod class changes:
* The new isIdempotent() and isHttpSafe() accessors are altered
slightly from the original code to match the RFCs. The existing code
produces incorrect true/false values on some methods.
* new accessor shouldInvalidate() which produces true if the request
method alone is enough to determine an RFC SHOULD invalidate condition.
- purgesOthers() accessor is still present with the old behaviour, but
for better caching we should convert the code to using
shouldInvalidate()
* new accessors respMaybeCacheable() which produces true if the
request method alone is enough to determine an RFC MAY cache condition.
- replaces isCacheable() accessor which did not produce any certainty
as implied by the name, and was producing incorrect assessments anyway
for some methods.
- it is more of a test whether *Squid* is known to be able to cache
the response to this method, than whether the RFC permit it.
- There are several cases where it should produce true according to
the RFCs but isCacheable() was doing false due to bugs or missing
feature support. These cases are now documented with reasons and things
to test or fix before they can be made to return true.
* the remainders of obsolete extension_methods directive code is dropped
Global httpCachable(method) function dropped - replaced by
HttpRequest::maybeCacheable() accessor which does better cacheability
checks.
... and there are a lot of METHOD_FOO -> Http::METHOD_FOO symbol-only
changes fluffing out the patch size.
Amos Jeffries [Tue, 23 Oct 2012 06:39:34 +0000 (19:39 +1300)]
Remove Bungled warning on missing component directives
* Make Squid display a ERROR notice and ignore squid.conf directives
which belong to components disabled at build time instead of throwing
up 'FATAL: Bungled' message and aborting.
* Polish the 'requires X' message text for better display of cases like
'requires MS Windows' and 'requires Linux'
* Add 'requires X' message for --enable-auth, --enable-eui, and
--enable-ssl-crtd which were missing.
Amos Jeffries [Mon, 22 Oct 2012 11:46:00 +0000 (00:46 +1300)]
Add constructore/destructor for Mgr::ActionPasswordList
Missing constructor for Mgr::ActionPasswordList resulted in invalid values
for next pointer and segmentation fault parsing multiple cachemgr_passwd
directives.
Missing destructor resulted in memory leaks on reconfigure and shutdown.
Amos Jeffries [Sat, 20 Oct 2012 08:01:32 +0000 (02:01 -0600)]
Remove the refresh_pattern ignore-no-cache option
With the HTTP/1.1 Cache-Control:no-cache compliance upgrade this option became
obsolete. Its main use-case was to force caching of responses which contained this
control. Revalidation is now happening and no-cache objects are cached by default.
The one remaining effect of ignoring no-cache is to prevent no-cache from allowing
authenticated responses being cached. Which is the opposite of its usually expected
result and allowing that change of behaviour would be counter productive.
Amos Jeffries [Fri, 19 Oct 2012 09:17:33 +0000 (03:17 -0600)]
HTTP/1.1: Cache-Control compliance upgrade
trunk rev 11361 converted Cache-Control header from using a single mask
bitmap (shared by request and response) to separate CC header objects in
the request response. This conversion contained several regressions like
the one bug 3670 reports.
This patch:
* documents HttpStateData::cacheableReply() clarifying the overall
method action and what each individual check it doing.
* resolves several visible regressions, including bug 3670.
* extends the caching to handle the "no-cache" controls as per HTTP/1.1
(MAY store, but MUST revalidate before use).
* extends the caching for several lesser known cases of "MAY store"
exemptions handling authenticated transactions.
* removes an old hack handling Pragme:no-cache (undefined in HTTP/1.1)
One side effect of now caching transactions utilizing "no-cache" is that
hacks around Pragma:no-cache are reduced to only having any effect when
Cache-Control is absent. Reducing their performance cost. And now require
--enable-http-violations is built in since HTTP/1.1 states that
response Pragma are not defined and thus SHOULD be ignored.
Amos Jeffries [Thu, 18 Oct 2012 08:43:32 +0000 (02:43 -0600)]
Windows: revert broken attempt at removing #define macro wrappers
"using Squid::foo" style of defining when namespace functions to use when
there is a clash appears not to work on MinGW. We are forced to retain
the old #define macros.
This resolves "error: call of overloaded ... is ambiguous"
Update a few functions parameters to POSIX compliant APIs using socklen_t.
This resolves "error: invalid conversion from 'socklen_t*'"
Amos Jeffries [Tue, 16 Oct 2012 01:38:43 +0000 (19:38 -0600)]
HTTP/1.1: honour Cache-Control before Pragma:no-cache
HTTPbis WG clarifications on the interaction between Cache-Control and
Pragma specify that Cache-Control may be explicitly crafted to allow
conditional caching by HTTP/1.1 caches and Pragma:no-cache supplied to
prevent caching by older HTTP/1.0 caches unable to understand the
Cache-Control properly.
Squid does obey Cache-Control. So we can deprecate Pragma to the side
cases when Cache-Control is absent entirely.
Alex Rousskov [Tue, 16 Oct 2012 00:26:06 +0000 (18:26 -0600)]
Reverted trunk r12255 changes. Provided a portable flexible arrays replacement.
Trunk r12255 made Clang compiler happy by removing flexible nonPod[] arrays.
Unfortunately, it also moved shared memory items into local memory (in some
cases uninitialized).
This change provides a Clang-friendly flexible array replacement while keeping
items in the shared memory (and using placement-new initialization). The code
may have become even less readable, but hopefully more portable.
N.B. Flexible arrays were introdiced in C99 standard, after C++ was
standardized in 1998. They are not yet in any revised C++ standard, but most
C++ compilers support them, at least for PODs.
Alex Rousskov [Tue, 16 Oct 2012 00:18:09 +0000 (18:18 -0600)]
Allow a ufs cache_dir entry to coexist with a shared memory cache entry
instead of being released when it becomes idle.
The original boolean version of the StoreController::dereference() code
(r11730) was written to make sure that idle unlocked local store_table entries
are released if nobody needs them (to avoid creating inconsistencies with
shared caches that could be modified in a different process).
Then, in r11786, we realized that the original code was destroying non-shared
memory cache entries if there were no cache_dirs to vote for keeping them in
store_table. I fixed that by changing the StoreController::dereference() logic
from "remove if nobody needs it" to "remove if somebody objects to keeping
it". That solved the problem at hand, but prohibited an entry to exist in
a non-shared cache_dir and in a shared memory cache at the same time.
We now go back to the original "remove if nobody needs it" design but also
give non-shared memory cache a vote so that it can protect idle but suitable
for memory cache entries from being released if there are no cache_dirs to
vote for them.
This is a second revision of the fix. The first one (r12231) was reverted
because it did not pass tests/testRock unit tests on some platforms. The unit
tests assume that the entry slot is not locked after the entry is stored, but
the first revision of the fix allowed idle entries to remain in store_table
and, hence, their slots were locked and could not be replaced, causing
assertions. This revision allows the idle entry to be destroyed (and its slot
unlocked) if [non-shared] memory caching is disabled.
It is not clear why only some of the platforms were affected by this. Should
not memory caching be disabled everywhere during testRock (because testRock
does not set memory cache capacity and memory cache entry size limits)?
Amos Jeffries [Fri, 12 Oct 2012 12:25:08 +0000 (06:25 -0600)]
Portability: pass OS name to helper config.test scripts
This will help us perform OS-specific enable/disable rather than having
to identify some software or file that SHOULD exist only in those OS, but
which might exist elsewhere for weird reasons (ie windows.h).
NP: the $@ which was being passed was unused by any current config.test
Removed it and converted to fixed-position parameters for easier
script development.
projects / thirdparty / squid.git / log
