]> git.ipfire.org Git - thirdparty/squid.git/log
thirdparty/squid.git
23 years agoBugfix: The aufs corruption patch triggered another bug in accounting
hno [Fri, 15 Nov 2002 20:16:31 +0000 (20:16 +0000)] 
Bugfix: The aufs corruption patch triggered another bug in accounting
of opening filedescriptors.

23 years agoBugzilla #463: GET requests with request entity fail
hno [Fri, 15 Nov 2002 20:12:36 +0000 (20:12 +0000)] 
Bugzilla #463: GET requests with request entity fail

This patch adds a new squid.conf directive "request_entities on/off".
If set to "on" then Squid will allow GET/HEAD requests with request
entities, even if such entites are "undefined" in the HTTP specification.

Technically speaking Squid is violating the HTTP specification when
not allowing such requests. The official way to handle such requests
is that the server should ignore the request entity.

23 years agoUse POST when submitting the login to prevent the login & password from
hno [Fri, 15 Nov 2002 20:10:18 +0000 (20:10 +0000)] 
Use POST when submitting the login to prevent the login & password from
being logged in the HTTP server access log or the browser history.

23 years agoSupport lines with spaces in them in included acl files
hno [Fri, 15 Nov 2002 20:09:31 +0000 (20:09 +0000)] 
Support lines with spaces in them in included acl files

23 years agoBugzilla #453: make uninstall should not remove squid.conf
hno [Fri, 15 Nov 2002 20:08:26 +0000 (20:08 +0000)] 
Bugzilla #453: make uninstall should not remove squid.conf

23 years agoCleanup of unused code warnings on --disable-memory-pools
hno [Fri, 15 Nov 2002 20:07:25 +0000 (20:07 +0000)] 
Cleanup of unused code warnings on --disable-memory-pools

23 years agodebug section bugfix
hno [Sun, 10 Nov 2002 11:43:04 +0000 (11:43 +0000)] 
debug section bugfix

23 years agoRemoved extra comma
hno [Sun, 10 Nov 2002 11:41:06 +0000 (11:41 +0000)] 
Removed extra comma

23 years agoC++ typecast fixes in WCCP
hno [Sun, 10 Nov 2002 11:37:14 +0000 (11:37 +0000)] 
C++ typecast fixes in WCCP

23 years agoIDENT C typecasts
hno [Sun, 10 Nov 2002 11:30:44 +0000 (11:30 +0000)] 
IDENT C typecasts

23 years agoMore IDENT C typecasts
hno [Sun, 10 Nov 2002 11:23:08 +0000 (11:23 +0000)] 
More IDENT C typecasts

23 years agoC typecasts in CARP code
hno [Sun, 10 Nov 2002 11:22:03 +0000 (11:22 +0000)] 
C typecasts in CARP code

23 years agoC typecasts in IDENT code
hno [Sun, 10 Nov 2002 11:19:39 +0000 (11:19 +0000)] 
C typecasts in IDENT code

23 years agoBootstrapped
hno [Sun, 10 Nov 2002 11:14:55 +0000 (11:14 +0000)] 
Bootstrapped

23 years agoFix --disable-... options to default to be enabled.. (CARP, WCCP, IDENT, ..)
hno [Sun, 10 Nov 2002 11:12:54 +0000 (11:12 +0000)] 
Fix --disable-... options to default to be enabled.. (CARP, WCCP, IDENT, ..)

23 years agoC++ requires type casting
hno [Sun, 10 Nov 2002 10:40:34 +0000 (10:40 +0000)] 
C++ requires type casting

23 years agodate: 2002/11/09 10:42:01; author: hno; state: Exp; lines: +46 -20
hno [Sun, 10 Nov 2002 09:29:58 +0000 (09:29 +0000)] 
date: 2002/11/09 10:42:01;  author: hno;  state: Exp;  lines: +46 -20
Bugzilla #451: Occasional corruption of objects

This patch fixes a cbdata boundary violation in the aufs storeio
implementation wich caused data corruption on aborted cache hits.

23 years agoGNUregex.h is a C header file. Needs to be wrapped in extern "C" when
hno [Sun, 10 Nov 2002 09:04:39 +0000 (09:04 +0000)] 
GNUregex.h is a C header file. Needs to be wrapped in extern "C" when
compiling C++ code.

23 years agoGet rid of compiler warnings on hash_lookup
hno [Sun, 10 Nov 2002 08:54:21 +0000 (08:54 +0000)] 
Get rid of compiler warnings on hash_lookup

23 years agoindent (from Squid-2)
hno [Sat, 9 Nov 2002 16:35:26 +0000 (16:35 +0000)] 
indent (from Squid-2)

23 years agodate: 2002/10/24 18:49:56; author: wessels; state: Exp; lines: +20 -4
hno [Sat, 9 Nov 2002 16:34:22 +0000 (16:34 +0000)] 
date: 2002/10/24 18:49:56;  author: wessels;  state: Exp;  lines: +20 -4
Bugzilla #462: WCCP doesn't ensure WCCP_ASSIGN_BUCKETS message was received

This patch makes Squid remember the change value when it last sent a
WCCP_ASSIGN_BUCKETS message.  It will resend the message until
the router updates the change value.

23 years agofix linux
robertc [Mon, 28 Oct 2002 18:27:47 +0000 (18:27 +0000)] 
fix linux

23 years agoWrap the memset() just before the msgrcv() in #ifdef ALWAYS_ZERO_BUFFERS.
adrian [Mon, 28 Oct 2002 15:53:37 +0000 (15:53 +0000)] 
Wrap the memset() just before the msgrcv() in #ifdef ALWAYS_ZERO_BUFFERS.
There's no real point in zeroing the buffer just before we read and then
verify the message is the right length - unless we're debugging.

23 years agomove the storeDirCallback() call out into main.cc where it
adrian [Mon, 28 Oct 2002 08:12:27 +0000 (08:12 +0000)] 
move the storeDirCallback() call out into main.cc where it
probably belongs.

23 years agoVarious modifications to make it work properly again.
adrian [Mon, 28 Oct 2002 07:37:36 +0000 (07:37 +0000)] 
Various modifications to make it work properly again.

23 years agoDamnit, only drop accept()s into a period event if we're UNDER the
adrian [Sun, 27 Oct 2002 21:19:39 +0000 (21:19 +0000)] 
Damnit, only drop accept()s into a period event if we're UNDER the
RESERVED_FD limit, not above!

23 years agoset a reasonable default for the accept checking.
adrian [Sun, 27 Oct 2002 21:15:49 +0000 (21:15 +0000)] 
set a reasonable default for the accept checking.

23 years agotypeo.
adrian [Sun, 27 Oct 2002 21:15:23 +0000 (21:15 +0000)] 
typeo.

23 years agoflesh out the accept defer processing magic.
adrian [Sun, 27 Oct 2002 21:13:35 +0000 (21:13 +0000)] 
flesh out the accept defer processing magic.
Its (hopefully!) usable now.

23 years agoCommit prototype
adrian [Sun, 27 Oct 2002 21:01:58 +0000 (21:01 +0000)] 
Commit prototype

23 years ago* whitespace
adrian [Sun, 27 Oct 2002 21:01:29 +0000 (21:01 +0000)] 
* whitespace
* add some debugging in comm_write() to see what we're doing
* add an fd entry to fdc_table[] which links back to the fd we are.
  This lets us pass &fdc_table[fd] to things like eventAdd when we're
  scheduling an event that is fd-related
* write a routine to set the accept poll delay - this will be used by
  the .. intelligent poll code. :)

23 years ago* remove the do {} while (1); crap - we don't need it here
adrian [Sun, 27 Oct 2002 20:45:06 +0000 (20:45 +0000)] 
* remove the do {} while (1); crap - we don't need it here
* fix the switch() statement and event handling - falling through is
  bad in this case as a registered event will only ever be a single
  type (unlike poll() where its a mask).

23 years agos/\t/ /g
adrian [Sun, 27 Oct 2002 20:06:49 +0000 (20:06 +0000)] 
s/\t/        /g

23 years agoremove unused variable.
adrian [Sat, 26 Oct 2002 10:57:29 +0000 (10:57 +0000)] 
remove unused variable.

23 years agofix incorrect underrun detection
robertc [Sat, 26 Oct 2002 10:51:01 +0000 (10:51 +0000)] 
fix incorrect underrun detection

23 years agoextract method and rename variable
robertc [Sat, 26 Oct 2002 08:42:31 +0000 (08:42 +0000)] 
extract method and rename variable

23 years agosimplify conditional
robertc [Sat, 26 Oct 2002 08:18:12 +0000 (08:18 +0000)] 
simplify conditional

23 years agoencapsulate tcp_reset flag
robertc [Fri, 25 Oct 2002 13:36:31 +0000 (13:36 +0000)] 
encapsulate tcp_reset flag

23 years agoRe-enable this assert, since I've fixed the pipe code.
adrian [Fri, 25 Oct 2002 11:54:44 +0000 (11:54 +0000)] 
Re-enable this assert, since I've fixed the pipe code.

23 years agofix mem leak if read fails
robertc [Fri, 25 Oct 2002 10:06:39 +0000 (10:06 +0000)] 
fix mem leak if read fails

23 years agooops
robertc [Fri, 25 Oct 2002 09:57:30 +0000 (09:57 +0000)] 
oops

23 years agobootstrap
robertc [Fri, 25 Oct 2002 09:48:56 +0000 (09:48 +0000)] 
bootstrap

23 years agofix a couple of mem leaks
robertc [Fri, 25 Oct 2002 09:47:03 +0000 (09:47 +0000)] 
fix a couple of mem leaks

23 years agocbdata history for cbdata debug mode
robertc [Fri, 25 Oct 2002 09:13:51 +0000 (09:13 +0000)] 
cbdata history for cbdata debug mode

23 years agomalloc_statistics() also needs to be defined with SQUIDCEXTERN!
adrian [Fri, 25 Oct 2002 07:25:26 +0000 (07:25 +0000)] 
malloc_statistics() also needs to be defined with SQUIDCEXTERN!

23 years agoExplicit type conversation, ta C++.
adrian [Fri, 25 Oct 2002 07:00:42 +0000 (07:00 +0000)] 
Explicit type conversation, ta C++.
(This was caught with xmalloc-stats no)

23 years agoAdd a cbdataReferenceDone() call to match the cbdataReference() call
adrian [Fri, 25 Oct 2002 05:22:29 +0000 (05:22 +0000)] 
Add a cbdataReferenceDone() call to match the cbdataReference() call
performed in aclChecklistCreate().

In a nutshell, this wasn't a memory leak, but it was a lock-leak -
the acl itself wasn't being freed, but the refcount slowly crept
upwards.

23 years agoSet an upper limit on lock counts (65535) - I'm sure we're about to trigger
adrian [Fri, 25 Oct 2002 04:59:29 +0000 (04:59 +0000)] 
Set an upper limit on lock counts (65535) - I'm sure we're about to trigger
it.

23 years agoAdd some more debugging ouput, helpful for tracking allocations in
adrian [Fri, 25 Oct 2002 04:53:44 +0000 (04:53 +0000)] 
Add some more debugging ouput, helpful for tracking allocations in
the cache.log file.

23 years ago* add in some more debugging statements
adrian [Fri, 25 Oct 2002 04:50:59 +0000 (04:50 +0000)] 
* add in some more debugging statements
* fix a bug where cbdataReferenceDone() wasn't being called on
  the callback data for a tcp connection. eep.

23 years agoFormatting/tidyups
adrian [Wed, 23 Oct 2002 17:24:20 +0000 (17:24 +0000)] 
Formatting/tidyups

23 years agowhen cancelling read events, also cancel the actual FD listen bit.
adrian [Wed, 23 Oct 2002 16:11:53 +0000 (16:11 +0000)] 
when cancelling read events, also cancel the actual FD listen bit.
(ta reub!)

23 years agoMake comm_listen/comm_accept .. optimistic.
adrian [Wed, 23 Oct 2002 16:11:03 +0000 (16:11 +0000)] 
Make comm_listen/comm_accept .. optimistic.
When we have an accept event we accept as many times as we can
(up until a #define'd limit) - if the callback doesn't re-register
we don't continue calling it :)

23 years agocomm_write() should operate on a const char *buf, not a char *buf
adrian [Wed, 23 Oct 2002 15:17:34 +0000 (15:17 +0000)] 
comm_write() should operate on a const char *buf, not a char *buf

23 years agotypo? rbuf[] -> buf[]
adrian [Wed, 23 Oct 2002 15:15:00 +0000 (15:15 +0000)] 
typo? rbuf[] -> buf[]

23 years agoMalformed author header comment fix by Guido
hno [Wed, 23 Oct 2002 03:16:11 +0000 (03:16 +0000)] 
Malformed author header comment fix by Guido

23 years agoicmp debug section cleanups by Guido
hno [Wed, 23 Oct 2002 03:12:51 +0000 (03:12 +0000)] 
icmp debug section cleanups by Guido

23 years agoBugfix: strwordtok parsing of quoted strings failed miserably if the last
hno [Tue, 22 Oct 2002 15:41:57 +0000 (15:41 +0000)] 
Bugfix: strwordtok parsing of quoted strings failed miserably if the last
word in was quoted.

23 years agoAdded a level 9 debug statement dumping the helper receive buffer
hno [Tue, 22 Oct 2002 15:40:29 +0000 (15:40 +0000)] 
Added a level 9 debug statement dumping the helper receive buffer
to ease debugging of helper interactions

23 years agoBootstrapped
hno [Tue, 22 Oct 2002 14:29:05 +0000 (14:29 +0000)] 
Bootstrapped

23 years agoEven more pthreads fixes. We have too much magics in pthreads detection.
hno [Tue, 22 Oct 2002 14:14:45 +0000 (14:14 +0000)] 
Even more pthreads fixes. We have too much magics in pthreads detection.

23 years agoMinor fix of the _REENTRANT issue on --enable-async-io. --enable-async-io
hno [Tue, 22 Oct 2002 14:07:03 +0000 (14:07 +0000)] 
Minor fix of the _REENTRANT issue on --enable-async-io. --enable-async-io
needs to trigger a --enable-pthreads, not just -D_REENTRANT.

23 years agoFix asyncio queue management when there is blocked requests.
hno [Tue, 22 Oct 2002 13:02:00 +0000 (13:02 +0000)] 
Fix asyncio queue management when there is blocked requests.

23 years agobootstrap
robertc [Tue, 22 Oct 2002 06:03:22 +0000 (06:03 +0000)] 
bootstrap

23 years agoset REENTRANT for pthreads, and do so for CXXFLAGS as well
robertc [Tue, 22 Oct 2002 05:53:22 +0000 (05:53 +0000)] 
set REENTRANT for pthreads, and do so for CXXFLAGS as well

23 years agoDebug section fixes by Guido
hno [Tue, 22 Oct 2002 04:17:32 +0000 (04:17 +0000)] 
Debug section fixes by Guido

23 years agos/comm_send/comm_udp_send/
hno [Mon, 21 Oct 2002 22:12:47 +0000 (22:12 +0000)] 
s/comm_send/comm_udp_send/

23 years agoconvert over a single (easy!) comm_old_write() call.
adrian [Mon, 21 Oct 2002 21:35:12 +0000 (21:35 +0000)] 
convert over a single (easy!) comm_old_write() call.

23 years ago* convert to new comm_write()
adrian [Mon, 21 Oct 2002 21:31:44 +0000 (21:31 +0000)] 
* convert to new comm_write()
* clear double-free buglet - memory was being freed both in
  gopherSendComplete() -and- comm_old_write() (via the free func)

23 years agocomm_write() conversion
adrian [Mon, 21 Oct 2002 21:26:09 +0000 (21:26 +0000)] 
comm_write() conversion

23 years agoconvert to use new comm_write().
adrian [Mon, 21 Oct 2002 21:21:52 +0000 (21:21 +0000)] 
convert to use new comm_write().

23 years agoMissed conversion.
adrian [Mon, 21 Oct 2002 21:18:32 +0000 (21:18 +0000)] 
Missed conversion.

23 years agoConvert to using the new comm_write()
adrian [Mon, 21 Oct 2002 21:13:40 +0000 (21:13 +0000)] 
Convert to using the new comm_write()

23 years agoFix typo which caused comm_write() to use the fill handler_data!
adrian [Mon, 21 Oct 2002 21:13:23 +0000 (21:13 +0000)] 
Fix typo which caused comm_write() to use the fill handler_data!

23 years agoChange some types. Damn C++.
adrian [Mon, 21 Oct 2002 20:48:15 +0000 (20:48 +0000)] 
Change some types. Damn C++.

23 years agoconvert comm_old_write() -> comm_write();
adrian [Mon, 21 Oct 2002 20:09:52 +0000 (20:09 +0000)] 
convert comm_old_write() -> comm_write();

23 years agoImplement a new comm_write().
adrian [Mon, 21 Oct 2002 20:08:22 +0000 (20:08 +0000)] 
Implement a new comm_write().

23 years ago* rename comm_write() -> comm_old_write()
adrian [Mon, 21 Oct 2002 20:00:01 +0000 (20:00 +0000)] 
* rename comm_write() -> comm_old_write()
* rename comm_write_mbuf() -> comm_old_write_mbuf()
* implement most of a replacement comm_write() call in comm.cc
  (everything except the actual comm_write() call is in there)

23 years agoNeeds #include "comm.h" for comm_udp_recvfrom()
hno [Mon, 21 Oct 2002 15:31:55 +0000 (15:31 +0000)] 
Needs #include "comm.h" for comm_udp_recvfrom()

23 years agorenaming.
adrian [Mon, 21 Oct 2002 15:18:40 +0000 (15:18 +0000)] 
renaming.

23 years agomove the udp functions to comm_udp_*, to be consistent with what was
adrian [Mon, 21 Oct 2002 12:43:07 +0000 (12:43 +0000)] 
move the udp functions to comm_udp_*, to be consistent with what was
in there in the first place.

23 years agosendto() -> comm_udp_sendto()
adrian [Mon, 21 Oct 2002 12:34:04 +0000 (12:34 +0000)] 
sendto() -> comm_udp_sendto()

23 years agosend() - > comm_send()
adrian [Mon, 21 Oct 2002 11:54:36 +0000 (11:54 +0000)] 
send() - > comm_send()

23 years agoadd a send(0 wrapper.
adrian [Mon, 21 Oct 2002 11:54:12 +0000 (11:54 +0000)] 
add a send(0 wrapper.

23 years agorecv -> comm_recv().
adrian [Mon, 21 Oct 2002 11:49:59 +0000 (11:49 +0000)] 
recv -> comm_recv().

23 years agoAdd in a recv() wrapper - comm_recv().
adrian [Mon, 21 Oct 2002 11:47:25 +0000 (11:47 +0000)] 
Add in a recv() wrapper - comm_recv().

From FreeBSD's recv/recvfrom:

     If from is non-nil, and the socket is not connection-oriented, the source
     address of the message is filled in.  Fromlen is a value-result parame-
     ter, initialized to the size of the buffer associated with from, and mod-
     ified on return to indicate the actual size of the address stored there.

.. so comm_recv() calls recv_from()

23 years agorecvfrom() -> comm_recvfrom()
adrian [Mon, 21 Oct 2002 11:45:30 +0000 (11:45 +0000)] 
recvfrom() -> comm_recvfrom()

23 years agoAdd in function decl.
adrian [Mon, 21 Oct 2002 11:44:55 +0000 (11:44 +0000)] 
Add in function decl.

23 years agoAdd in a recvfrom() wrapper - its still non-callback, but it means I can
adrian [Mon, 21 Oct 2002 11:38:32 +0000 (11:38 +0000)] 
Add in a recvfrom() wrapper - its still non-callback, but it means I can
tidy things up and remove the direct socket calls from other places.

23 years agoMake temporary backup have .bak extension
hno [Mon, 21 Oct 2002 05:19:09 +0000 (05:19 +0000)] 
Make temporary backup have .bak extension

23 years agoBootstrapped
hno [Mon, 21 Oct 2002 05:12:50 +0000 (05:12 +0000)] 
Bootstrapped

23 years agoOnly invoke the automake workaround if needed
hno [Mon, 21 Oct 2002 05:08:44 +0000 (05:08 +0000)] 
Only invoke the automake workaround if needed

23 years agoBootstrapped
hno [Sun, 20 Oct 2002 06:10:49 +0000 (06:10 +0000)] 
Bootstrapped

23 years agoFix a small keyboard typo / vi ghost..
hno [Sun, 20 Oct 2002 03:50:22 +0000 (03:50 +0000)] 
Fix a small keyboard typo / vi ghost..

23 years agoSquid releasenotes snapshots (merged from SQUID2)
hno [Sat, 19 Oct 2002 18:29:00 +0000 (18:29 +0000)] 
Squid releasenotes snapshots (merged from SQUID2)

23 years agoadd more non listed files
robertc [Sat, 19 Oct 2002 08:05:31 +0000 (08:05 +0000)] 
add more non listed files

23 years agosupport --removal-policies again, and add Store.h to the listed source files
robertc [Sat, 19 Oct 2002 07:23:21 +0000 (07:23 +0000)] 
support --removal-policies again, and add Store.h to the listed source files

23 years agoremove repl makefiles again, dist check works fine now
robertc [Sat, 19 Oct 2002 07:00:25 +0000 (07:00 +0000)] 
remove repl makefiles again, dist check works fine now

23 years agoIndent (merged from Squid-2)
hno [Sat, 19 Oct 2002 04:46:46 +0000 (04:46 +0000)] 
Indent (merged from Squid-2)

23 years agodate: 2002/10/17 04:54:30; author: wessels; state: Exp; lines: +17 -16
hno [Sat, 19 Oct 2002 04:45:44 +0000 (04:45 +0000)] 
date: 2002/10/17 04:54:30;  author: wessels;  state: Exp;  lines: +17 -16
use 'hash_link hash' trick instead of 'first two items must be same
as hash_link' technique.

format comments for 80 cols.
(Merged from Squid-2)