]> git.ipfire.org Git - thirdparty/squid.git/log
thirdparty/squid.git
26 years agoinvented snmp_var_new_integer() because I think snmp memory allocation belongs in...
wessels [Fri, 18 Jun 1999 04:20:35 +0000 (04:20 +0000)] 
invented snmp_var_new_integer() because I think snmp memory allocation belongs in snmplib

26 years agoUse MEM_SNMP_SNINT type for allocating SNMP reply stuff
wessels [Fri, 18 Jun 1999 02:23:08 +0000 (02:23 +0000)] 
Use MEM_SNMP_SNINT type for allocating SNMP reply stuff

26 years agoreword -N usage description
wessels [Thu, 17 Jun 1999 11:02:21 +0000 (11:02 +0000)] 
reword -N usage description

26 years agoset reply->hdr_sz for these buggers so store.log makes sense
wessels [Thu, 17 Jun 1999 10:04:04 +0000 (10:04 +0000)] 
set reply->hdr_sz for these buggers so store.log makes sense

26 years agoAdded maximum_icp_query_timeout option because John Line sometimes
wessels [Thu, 17 Jun 1999 04:10:36 +0000 (04:10 +0000)] 
Added maximum_icp_query_timeout option because John Line sometimes
sees dynamic query timeouts in the range of 6-7 seconds.

26 years agoAdded redirector_bypass option, written for Deere and Company
wessels [Thu, 17 Jun 1999 03:23:31 +0000 (03:23 +0000)] 
Added redirector_bypass option, written for Deere and Company

26 years agoonoff.persistent_client_posts is old and unused
wessels [Thu, 17 Jun 1999 03:22:14 +0000 (03:22 +0000)] 
onoff.persistent_client_posts is old and unused

26 years agoIgnore an ICP message if the ICP length is not equal to the number
wessels [Mon, 14 Jun 1999 09:02:06 +0000 (09:02 +0000)] 
Ignore an ICP message if the ICP length is not equal to the number
of bytes read.

26 years agoGRE code removal, allow it to be done in the kernel
glenn [Sat, 12 Jun 1999 05:30:36 +0000 (05:30 +0000)] 
GRE code removal, allow it to be done in the kernel

26 years ago2.2 branch merge
wessels [Sat, 12 Jun 1999 04:40:32 +0000 (04:40 +0000)] 
2.2 branch merge

26 years agohierarchy logging was broken for SSL
wessels [Fri, 11 Jun 1999 03:06:23 +0000 (03:06 +0000)] 
hierarchy logging was broken for SSL

26 years agoadding
wessels [Fri, 11 Jun 1999 02:18:22 +0000 (02:18 +0000)] 
adding

26 years agoFinally finished the bit of HTCP that checks the query HTTP headers
wessels [Thu, 10 Jun 1999 12:10:30 +0000 (12:10 +0000)] 
Finally finished the bit of HTCP that checks the query HTTP headers
for freshness.

26 years ago- the max-age cc option may be preceded with a comma like any other option
rousskov [Mon, 31 May 1999 11:03:35 +0000 (11:03 +0000)] 
- the max-age cc option may be preceded with a comma like any other option

26 years agowhy did I remove Makefile.in in distclean target??
wessels [Thu, 27 May 1999 11:56:31 +0000 (11:56 +0000)] 
why did I remove Makefile.in in distclean target??

26 years agoFreeBSD's pthreads breaks dup2(). This means we can't use dup2()
wessels [Thu, 27 May 1999 09:26:10 +0000 (09:26 +0000)] 
FreeBSD's pthreads breaks dup2().  This means we can't use dup2()
to try to find SQUID_MAXFD in configure.  To get around the bug,
we make a temporary LDFLAGS string and remove -pthread.

26 years agoprogress on async-UFS code
wessels [Thu, 27 May 1999 09:21:37 +0000 (09:21 +0000)] 
progress on async-UFS code

26 years agochange frequency of validated-so-far output
wessels [Thu, 27 May 1999 09:13:12 +0000 (09:13 +0000)] 
change frequency of validated-so-far output

26 years agocompile bugs
wessels [Thu, 27 May 1999 08:42:31 +0000 (08:42 +0000)] 
compile bugs

26 years agocommitting first stab at async disk IO with new disk storage interface
wessels [Thu, 27 May 1999 02:41:15 +0000 (02:41 +0000)] 
committing first stab at async disk IO with new disk storage interface

26 years agofile_open() is entirely blocking now -- there is no need for the
wessels [Wed, 26 May 1999 23:07:56 +0000 (23:07 +0000)] 
file_open() is entirely blocking now -- there is no need for the
callback function arguments.

26 years agorename SwapDir->obj.log to SwapDir->log.write
wessels [Wed, 26 May 1999 12:48:06 +0000 (12:48 +0000)] 
rename SwapDir->obj.log to SwapDir->log.write
storeUfsFilenoBelongsHere() can be static

26 years agocleanup after actually reading it
wessels [Wed, 26 May 1999 12:40:43 +0000 (12:40 +0000)] 
cleanup after actually reading it

26 years agoadded big store system chunk
wessels [Wed, 26 May 1999 12:24:21 +0000 (12:24 +0000)] 
added big store system chunk

26 years agoSome fixes for swapping out objects that are in STORE_OK state. Need
wessels [Wed, 26 May 1999 11:01:05 +0000 (11:01 +0000)] 
Some fixes for swapping out objects that are in STORE_OK state.  Need
to make sure that we really write all the data and queue the
swapout file for closing.

DO NOT allow swapout of non-special objects during store rebuilding
phase.  This greatly speeds up the rebuild process and prevents
large memory consumption from objects that want to be deleted but
must wait until validation completes.

26 years agoAnother while loop in storeCleanup() event so that we do at least
wessels [Wed, 26 May 1999 10:58:52 +0000 (10:58 +0000)] 
Another while loop in storeCleanup() event so that we do at least
50 entries per call, instead of just one hash bucket which might
only have a handful of objects.

26 years agoBuffer the StoreEntry when building a mime icon object.
wessels [Wed, 26 May 1999 10:57:46 +0000 (10:57 +0000)] 
Buffer the StoreEntry when building a mime icon object.
Also set the ENTRY_SPECIAL flag before calling storeComplete so
that it gets swapped out with the new no-swapout-during-rebuild
scheme.

26 years agolooks like storeEntryDump 'e' arg can be const
wessels [Wed, 26 May 1999 10:36:55 +0000 (10:36 +0000)] 
looks like storeEntryDump 'e' arg can be const

26 years agoSTORE_SWAP_BUF and VM_WINDOW_SZ are the same as DISK_PAGE_SIZE. Having
wessels [Wed, 26 May 1999 09:18:15 +0000 (09:18 +0000)] 
STORE_SWAP_BUF and VM_WINDOW_SZ are the same as DISK_PAGE_SIZE.  Having
different names is only confusing.

26 years agoTwo storeSwapOut() changes:
wessels [Wed, 26 May 1999 09:08:12 +0000 (09:08 +0000)] 
Two storeSwapOut() changes:

1) with new store-FS scheme, we don't get write callbacks.  This means
we were getting some objects "stuck" because we had, say 32K in memory,
but only wrote 8K in storeSwapOut() and there was no signal to make
us write another block.  Now we write as many blocks as we can,
instead of just one.

2) Re-fixed the case where storeSwapOut() gets called because the
store_status has changed to STORE_OK, but there is no data to swap
out.  Now close the swapout file, as done at the end of storeSwapOut().

26 years agoMoved ->fd to UFS specific section of storeIOState structure
wessels [Wed, 26 May 1999 04:18:20 +0000 (04:18 +0000)] 
Moved ->fd to UFS specific section of storeIOState structure

26 years agouse doubles for higher precision elapsed time calculations
wessels [Wed, 26 May 1999 04:10:13 +0000 (04:10 +0000)] 
use doubles for higher precision elapsed time calculations

26 years agoMoved storeDirClean() into UFS-specific module.
wessels [Wed, 26 May 1999 04:05:47 +0000 (04:05 +0000)] 
Moved storeDirClean() into UFS-specific module.
random other ufs/api mods

26 years agowrong function argument
wessels [Wed, 26 May 1999 02:38:06 +0000 (02:38 +0000)] 
wrong function argument

26 years agoforgot to decrement store_dirs_rebuilding in storeCleanup()
wessels [Tue, 25 May 1999 13:02:00 +0000 (13:02 +0000)] 
forgot to decrement store_dirs_rebuilding in storeCleanup()

26 years agomassive store filesystem reorganization. Removing hard-coded hooks
wessels [Tue, 25 May 1999 12:53:40 +0000 (12:53 +0000)] 
massive store filesystem reorganization.  Removing hard-coded hooks
for UFS storage and using function pointers instead.  Yummmm...

26 years agoContinuing the process of supporting multiple filesystem types. Now
wessels [Sat, 22 May 1999 13:42:01 +0000 (13:42 +0000)] 
Continuing the process of supporting multiple filesystem types.  Now
the I/O operations are function pointers in the SwapDir structure.

Moved UFS-specific parsing out of cache_cf.c and into store_dir_ufs.c.
Then had to make some cache_cf.c functions global.

Removed some remaining bits of USE_ASYNC_IO code.

26 years agoFixed a bug with closing swapout files around the time we call
wessels [Sat, 22 May 1999 08:31:18 +0000 (08:31 +0000)] 
Fixed a bug with closing swapout files around the time we call
storeComplete().  Used to be that storeComplete() would call
storeClose() if there was a swapout.sio structure.  This was
essentially the right thing to do, but kind of awkwardly implemented.

A better choice is to have storeSwapOut() close objects that are
in STORE_OK state and if inmem_hi == queued_offset.

26 years agoAdded log_type to 'active_requests' output
wessels [Sat, 22 May 1999 08:28:56 +0000 (08:28 +0000)] 
Added log_type to 'active_requests' output

26 years agodocument storeIOState->offset
wessels [Sat, 22 May 1999 05:34:11 +0000 (05:34 +0000)] 
document storeIOState->offset

26 years agohandle the case when storeOpen() fails.
wessels [Sat, 22 May 1999 04:23:25 +0000 (04:23 +0000)] 
handle the case when storeOpen() fails.

26 years agonull pointer access on rep
wessels [Sat, 22 May 1999 04:16:56 +0000 (04:16 +0000)] 
null pointer access on rep

26 years agosgml compiler
wessels [Thu, 20 May 1999 04:47:53 +0000 (04:47 +0000)] 
sgml compiler

26 years agogrammar
wessels [Thu, 20 May 1999 04:46:19 +0000 (04:46 +0000)] 
grammar

26 years agogrammar
wessels [Thu, 20 May 1999 04:38:18 +0000 (04:38 +0000)] 
grammar

26 years agodon't need error flag on STRCB because we'll call STIOCB on error
wessels [Thu, 20 May 1999 04:36:42 +0000 (04:36 +0000)] 
don't need error flag on STRCB because we'll call STIOCB on error
and STRCB is only called upon a successful read.

26 years agotweak
wessels [Thu, 20 May 1999 04:35:46 +0000 (04:35 +0000)] 
tweak

26 years agoadded store filesystem docs
wessels [Thu, 20 May 1999 04:33:36 +0000 (04:33 +0000)] 
added store filesystem docs

26 years agoSince there is no write callback with new store filesystem interface,
wessels [Thu, 20 May 1999 03:57:47 +0000 (03:57 +0000)] 
Since there is no write callback with new store filesystem interface,
we must pass a free function to the low level write routines to free
the write buffer after its done

26 years agodnsStats only needed when using dnsservers
wessels [Thu, 20 May 1999 03:55:51 +0000 (03:55 +0000)] 
dnsStats only needed when using dnsservers

26 years agoyucky formatting
wessels [Thu, 20 May 1999 02:48:47 +0000 (02:48 +0000)] 
yucky formatting

26 years agoupdate
wessels [Thu, 20 May 1999 02:30:41 +0000 (02:30 +0000)] 
update

26 years ago2.2 branch merge
wessels [Thu, 20 May 1999 02:15:45 +0000 (02:15 +0000)] 
2.2 branch merge

26 years ago2.2 branch merge
wessels [Thu, 20 May 1999 01:57:34 +0000 (01:57 +0000)] 
2.2 branch merge

26 years agocomments don't match default
wessels [Wed, 12 May 1999 22:35:57 +0000 (22:35 +0000)] 
comments don't match default

26 years agoMarkus Leutwyler wrote:
wessels [Wed, 12 May 1999 02:14:14 +0000 (02:14 +0000)] 
Markus Leutwyler wrote:
>I have attached two logfiles (cache.log .. taken with debug = all,9)
>that show the different behaviour between the small text file and the
>bigger binary file. The FTP server is replying to the SIZE command with
>"213 0" (0 bytes) and this is sent to the browser as Content-length. I
>just wonder why the text file works and the binary file not. Could
>someone look at the logfile and try to figure out what's going on
>exactly?

Henrik:
SIZE is only used on binary transfers, not text transfers.  Attached is
a small workaround which causes Squid to ignore 0 replies to SIZE.

26 years agoKenneth R. Brownfield wrote:
wessels [Wed, 12 May 1999 02:11:48 +0000 (02:11 +0000)] 
Kenneth R. Brownfield wrote:
> When squid's configure was modified to turn on dlmalloc automatically for
> Solaris/x86, it looks like it didn't get turned on completely.
> Running configure without --enable-dlmalloc says that it turns on dlmalloc
> by default, but 'make' never compiles or links with dlmalloc.
> --enable-dlmalloc is still necessary to truly include dlmalloc.

From:    Henrik Nordstrom <hno@hem.passagen.se>
Patch attached for the dlmalloc configure issue.

26 years agoFixed a bug with the new ERR_TOO_BIG error message. Somtimes
wessels [Wed, 12 May 1999 00:51:47 +0000 (00:51 +0000)] 
Fixed a bug with the new ERR_TOO_BIG error message.  Somtimes
we'll get non-GET requests that don't have content-length
(like CONNECT, for example).

Also tried to force closing of persistent connections after sending
an "unsupported request" error message.  Those mostly come from
buggy browsers that send SSL requests down a persistent HTTP connection.

26 years agoIn function neighborsDigestSelect(): added a check for the
wessels [Wed, 12 May 1999 00:45:46 +0000 (00:45 +0000)] 
In function neighborsDigestSelect(): added a check for the
request->flag.hierarchical flag.  If the flag is not set, then
we should not use cache digests to select a neighbor.

Normally this was not a problem because non-hierarchical
requests are usually also non-cachable, and digests only
include cachable objects.  However, it seems that we sometimes
get false hits from the digest for non-hierarchical objects,
specifically a POST request.

If the POST goes to a sibling, we'll add only-if-cached.  Of course
the POST response should not be cached, so the response is 504
Gateway Timeout.  The originating cache can not re-forward the
request because it was a POST (well, maybe it could, but thats
a slippery slope).  The user gets the ugly cache-miss and
only-if-cached error message.

26 years agomissed another spot to use ERR_TOO_BIG
wessels [Tue, 11 May 1999 01:45:55 +0000 (01:45 +0000)] 
missed another spot to use ERR_TOO_BIG

26 years agoDW applied ERR_TOO_BIG to the wrong error message
wessels [Tue, 11 May 1999 01:43:36 +0000 (01:43 +0000)] 
DW applied ERR_TOO_BIG to the wrong error message

26 years agoAdded configuration parameters for maximum sizes of request headers,
wessels [Tue, 11 May 1999 01:33:20 +0000 (01:33 +0000)] 
Added configuration parameters for maximum sizes of request headers,
request bodies, and reply bodies.  Added new error page (ERR_TOO_BIG).

Large request bodies can be detected immediately because the request
header must have a valid content length.

Large reply bodies can be detected either by the content-length reply
header, or if not present, by the number of bytes written to the
client.

26 years agofixed potential type comparison bug
wessels [Tue, 11 May 1999 01:27:50 +0000 (01:27 +0000)] 
fixed potential type comparison bug

26 years agoadded ERR_TOO_BIG error message
wessels [Tue, 11 May 1999 01:25:21 +0000 (01:25 +0000)] 
added ERR_TOO_BIG error message

26 years ago2.2 branch merge
wessels [Wed, 5 May 1999 03:58:18 +0000 (03:58 +0000)] 
2.2 branch merge

26 years agogindent and 2.2 branch merge
wessels [Wed, 5 May 1999 03:20:36 +0000 (03:20 +0000)] 
gindent and 2.2 branch merge

26 years agoupdate
wessels [Wed, 5 May 1999 03:10:37 +0000 (03:10 +0000)] 
update

26 years agoaccessing request->user_ident was unguarded. It should be moved inside
wessels [Wed, 5 May 1999 03:08:40 +0000 (03:08 +0000)] 
accessing request->user_ident was unguarded.  It should be moved inside
the if (request) statement
Luyer

26 years agoremove Makefile.in.bak, which might get created by makedepend
wessels [Wed, 5 May 1999 03:06:21 +0000 (03:06 +0000)] 
remove Makefile.in.bak, which might get created by makedepend
Luyer

26 years agoOn make distclean, remove makefile backup (made by make depend if it is used)
wessels [Wed, 5 May 1999 03:01:04 +0000 (03:01 +0000)] 
On make distclean, remove makefile backup (made by make depend if it is used)
in lib/.
Luyer

26 years agoFunctions checked for in AC_REPLACE_FUNCS don't need to be checked for in
wessels [Wed, 5 May 1999 03:00:34 +0000 (03:00 +0000)] 
Functions checked for in AC_REPLACE_FUNCS don't need to be checked for in
AC_CHECK_FUNCS (re-run autoconf after applying).
Luyer

26 years agoundefined function from henriks patch
wessels [Wed, 5 May 1999 02:58:31 +0000 (02:58 +0000)] 
undefined function from henriks patch

26 years agoAdded peer_connect_timeout option
wessels [Wed, 5 May 1999 02:49:36 +0000 (02:49 +0000)] 
Added peer_connect_timeout option

added connect-timeout option too cache_peer directive

26 years agohenrik patch to handle multiline headers
wessels [Wed, 5 May 1999 01:43:17 +0000 (01:43 +0000)] 
henrik patch to handle multiline headers

26 years agoFrom: Alex Rousskov <rousskov@ircache.net>
wessels [Wed, 5 May 1999 01:39:00 +0000 (01:39 +0000)] 
From:    Alex Rousskov <rousskov@ircache.net>

Here is a small patch to make Squid more robust if cache or server
clock is skewed. The right thing is probably to make age and
entry->timestamp semi-independant by adding a function that computes an
age based on all info available, including entry->timestamp. The patch
below simply adjusts entry->timestamp so it does not lead to negative
ages later.  If entry->timestamp should be based on "cache time" rather
than "server time", the patch should not have bad side effects.

26 years agoadd FD to debugging message
wessels [Wed, 5 May 1999 01:26:45 +0000 (01:26 +0000)] 
add FD to debugging message

26 years agogindent
wessels [Wed, 5 May 1999 01:22:21 +0000 (01:22 +0000)] 
gindent

26 years agodont need statvfs here
wessels [Wed, 5 May 1999 01:14:27 +0000 (01:14 +0000)] 
dont need statvfs here

26 years agoprintf bug
wessels [Wed, 5 May 1999 00:47:28 +0000 (00:47 +0000)] 
printf bug

26 years agobad domain matching diagnostic
wessels [Tue, 4 May 1999 23:40:09 +0000 (23:40 +0000)] 
bad domain matching diagnostic

26 years agomove some store initialization stuff into storeUfsDirInit()
wessels [Tue, 4 May 1999 04:59:56 +0000 (04:59 +0000)] 
move some store initialization stuff into storeUfsDirInit()

26 years agoremove silly message about replacement rate
wessels [Tue, 4 May 1999 04:43:34 +0000 (04:43 +0000)] 
remove silly message about replacement rate

26 years agosplit a bunch of FS-dependent stuff from store_dir.c into store_dir_ufs.c,
wessels [Tue, 4 May 1999 04:41:09 +0000 (04:41 +0000)] 
split a bunch of FS-dependent stuff from store_dir.c into store_dir_ufs.c,
but its probably not really enough.

26 years agoAn initial implementation of new store API. Everything goes through
wessels [Tue, 4 May 1999 03:54:55 +0000 (03:54 +0000)] 
An initial implementation of new store API.  Everything goes through
fucntions in store_io.c, which is basically a level of indirection
for different types of filesystems.

This still needs a lot of work.  Have to move all rebuild functions
into store_io_ufs.c.  Have to add the indirection, instead of hard
coding everything to the UFS code.

26 years agoremove OPTIMISTIC_IO crap
wessels [Tue, 4 May 1999 02:39:30 +0000 (02:39 +0000)] 
remove OPTIMISTIC_IO crap

26 years agofix refresh stats
wessels [Tue, 4 May 1999 02:37:49 +0000 (02:37 +0000)] 
fix refresh stats
   - show 0 entries too
   - put a new line after "total" and before FYI entries

26 years agoremoved leftover? assertion from debugging days
wessels [Mon, 3 May 1999 22:29:45 +0000 (22:29 +0000)] 
removed leftover? assertion from debugging days

26 years agotypo
wessels [Sat, 1 May 1999 00:06:34 +0000 (00:06 +0000)] 
typo

26 years agoupdate from Jedi/Sector One
wessels [Wed, 28 Apr 1999 04:22:02 +0000 (04:22 +0000)] 
update from Jedi/Sector One

26 years agoluyer nitpick
wessels [Wed, 28 Apr 1999 04:11:28 +0000 (04:11 +0000)] 
luyer nitpick

26 years agoupdate from Bertold Kolics
wessels [Wed, 28 Apr 1999 00:50:17 +0000 (00:50 +0000)] 
update from Bertold Kolics

26 years agoa bit more friendly debugging to tell you if a request was allowed
wessels [Tue, 27 Apr 1999 12:33:38 +0000 (12:33 +0000)] 
a bit more friendly debugging to tell you if a request was allowed
or denied, and by which ACL

26 years agoGot GRE working
glenn [Tue, 27 Apr 1999 11:46:13 +0000 (11:46 +0000)] 
Got GRE working

26 years agoonly set the DNS socket read handler when we have outstanding queries,
wessels [Tue, 27 Apr 1999 04:34:06 +0000 (04:34 +0000)] 
only set the DNS socket read handler when we have outstanding queries,
instead of having it set all the time

26 years agorearrange so we can actually FIND comm_poll_dns_incoming
wessels [Tue, 27 Apr 1999 04:32:15 +0000 (04:32 +0000)] 
rearrange so we can actually FIND comm_poll_dns_incoming

26 years agoDeleted unused SNMP declaration
glenn [Tue, 27 Apr 1999 03:52:39 +0000 (03:52 +0000)] 
Deleted unused SNMP declaration

26 years agopreprocessor typo
wessels [Tue, 27 Apr 1999 03:52:17 +0000 (03:52 +0000)] 
preprocessor typo

26 years agomoved #define SQUID_NONBLOCK if-statement to src/squid.h because
wessels [Tue, 27 Apr 1999 03:36:02 +0000 (03:36 +0000)] 
moved #define SQUID_NONBLOCK if-statement to src/squid.h because
O_NONBLOCK probably won't be defined when config.h is read!

26 years agobetter debug message
wessels [Tue, 27 Apr 1999 03:09:52 +0000 (03:09 +0000)] 
better debug message

26 years agoadded httpReplyBodySize() -- returns the size of a reply message body.
wessels [Tue, 27 Apr 1999 03:06:12 +0000 (03:06 +0000)] 
added httpReplyBodySize() -- returns the size of a reply message body.
Use it in http.c and client_side.c