]>
git.ipfire.org Git - thirdparty/squid.git/log
wessels [Sat, 25 Apr 1998 03:34:07 +0000 (03:34 +0000)]
update
wessels [Sat, 25 Apr 1998 03:33:31 +0000 (03:33 +0000)]
adding Italian errors
wessels [Sat, 25 Apr 1998 00:48:08 +0000 (00:48 +0000)]
update
wessels [Sat, 25 Apr 1998 00:47:42 +0000 (00:47 +0000)]
Removed hacks in peer_select.c for ASN lookups. We cannot delay
wait for IP lookups in the peer selection process, and we do not
want to lookup EVERY address there. Instead, the acl.c routines
will lookup IP addresses for hostnames if they are not available.
So the first time we check some host, it may not match, but should
be present for subsequent requests.
wessels [Sat, 25 Apr 1998 00:35:59 +0000 (00:35 +0000)]
update
wessels [Sat, 25 Apr 1998 00:35:46 +0000 (00:35 +0000)]
update
rousskov [Sat, 25 Apr 1998 00:32:11 +0000 (00:32 +0000)]
- fixed comments after merging two DIGEST #defines
rousskov [Sat, 25 Apr 1998 00:30:03 +0000 (00:30 +0000)]
- added a few b20 changes
wessels [Sat, 25 Apr 1998 00:27:50 +0000 (00:27 +0000)]
memInit is required for config parsing
wessels [Fri, 24 Apr 1998 13:29:37 +0000 (13:29 +0000)]
typo
wessels [Fri, 24 Apr 1998 13:09:26 +0000 (13:09 +0000)]
gindent
wessels [Fri, 24 Apr 1998 13:08:39 +0000 (13:08 +0000)]
update
wessels [Fri, 24 Apr 1998 12:40:05 +0000 (12:40 +0000)]
remove wierd , CUR in ipcache output
wessels [Fri, 24 Apr 1998 12:37:23 +0000 (12:37 +0000)]
Henrik:
Here is a patch to tell configure to search for regex_t, and if
not found compile gnu regex.
wessels [Fri, 24 Apr 1998 12:22:44 +0000 (12:22 +0000)]
update
wessels [Fri, 24 Apr 1998 12:08:13 +0000 (12:08 +0000)]
unifiy DIGEST #ifdefs into USE_CACHE_DIGESTS
wessels [Fri, 24 Apr 1998 12:00:01 +0000 (12:00 +0000)]
fix WARNINGs
wessels [Fri, 24 Apr 1998 11:54:19 +0000 (11:54 +0000)]
henrik ftp fixes
wessels [Fri, 24 Apr 1998 11:44:08 +0000 (11:44 +0000)]
Add totals for ICP and HTTP hits
wessels [Fri, 24 Apr 1998 11:26:39 +0000 (11:26 +0000)]
configurable language
wessels [Fri, 24 Apr 1998 11:26:17 +0000 (11:26 +0000)]
update
wessels [Fri, 24 Apr 1998 11:24:43 +0000 (11:24 +0000)]
adding
wessels [Fri, 24 Apr 1998 11:20:23 +0000 (11:20 +0000)]
windows NT defines _timezone apparently
wessels [Fri, 24 Apr 1998 11:05:10 +0000 (11:05 +0000)]
Henrik:
* Remember the client address on redirected requests
* Don't rebuild the request if the redirector returned the same URL
* Rewrite Host: header on redirected requests
* Include port (if non-standard) in generated Host: headers
wessels [Fri, 24 Apr 1998 10:52:40 +0000 (10:52 +0000)]
fix urlInternal; no port number if its the default HTTP port
wessels [Fri, 24 Apr 1998 10:52:07 +0000 (10:52 +0000)]
comment
wessels [Fri, 24 Apr 1998 10:23:07 +0000 (10:23 +0000)]
ifdef http->lookup_type
wessels [Fri, 24 Apr 1998 04:51:04 +0000 (04:51 +0000)]
Fix ip acls with SPLAY trees
wessels [Fri, 24 Apr 1998 04:30:19 +0000 (04:30 +0000)]
vpath fix
rousskov [Fri, 24 Apr 1998 04:04:47 +0000 (04:04 +0000)]
- more "huge 304 reply" fixes
rousskov [Fri, 24 Apr 1998 02:25:51 +0000 (02:25 +0000)]
- added storeUnlockObject() to the "huge 304 reply" fix
- removed STORE_ABORTED if in the same section since it cannot be reached
wessels [Fri, 24 Apr 1998 02:11:09 +0000 (02:11 +0000)]
Fixed snprintf() off-by-one bug and found out that snprintf()
might emulate sprintf() behaviour with len == 0 args.
Now. we assert() that sprintf() length is >= 0. If len == 0, then
strx_printv() silently returns, leaving the buffer untouched.
rousskov [Fri, 24 Apr 1998 02:08:07 +0000 (02:08 +0000)]
- untested fix for huge 304 replies; if 304 reply is bigger than 4K we do not
have enough buffers to process it. Use the old entry instead.
wessels [Fri, 24 Apr 1998 01:29:12 +0000 (01:29 +0000)]
Changed peerCheckConnect() to use cbdata instead of flag to know
when to delete events
wessels [Fri, 24 Apr 1998 01:28:13 +0000 (01:28 +0000)]
wordlistCat
wessels [Fri, 24 Apr 1998 01:27:22 +0000 (01:27 +0000)]
Fixed buffer overruns from FTP server messages in errorConvert(). Now use
wordlistCat() which checks lengths *before* calling snprintf()
rousskov [Thu, 23 Apr 1998 09:21:16 +0000 (09:21 +0000)]
- check version requirements _both_ ways; "ignore digest" conditions:
had: cblock.ver.required > CacheDigestVer.current
added: cblock.ver.current < CacheDigestVer.required
rousskov [Thu, 23 Apr 1998 03:28:24 +0000 (03:28 +0000)]
- synced with current CacheDigest interface
rousskov [Wed, 22 Apr 1998 22:25:15 +0000 (22:25 +0000)]
- added/removed prototypes related to the second edition of Cache Digests
rousskov [Wed, 22 Apr 1998 22:24:12 +0000 (22:24 +0000)]
- polish
rousskov [Wed, 22 Apr 1998 22:23:41 +0000 (22:23 +0000)]
- temporary support for X-Cache-Lookup header field/stats
rousskov [Wed, 22 Apr 1998 22:23:20 +0000 (22:23 +0000)]
- new fields for digest cblock
- increased reserved area of cblock
- temporary support for X-Cache-Lookup header field/stats
rousskov [Wed, 22 Apr 1998 22:22:58 +0000 (22:22 +0000)]
- better synchronization between Rebuild and Rewrite
now each process can wait for other to complete
- added storeDigestAdd/Del() to check object freshness
- better, but still not perfect, pre-calculation of
cache digest size
- stats for additions, deletions, collisions, etc.
- polish
rousskov [Wed, 22 Apr 1998 22:22:37 +0000 (22:22 +0000)]
- removed calls to cacheDigestAdd/cacheDigestDel()
soon we will cancel support of in-memory copy of
a local cache digest
rousskov [Wed, 22 Apr 1998 22:22:15 +0000 (22:22 +0000)]
- moved Config.Refresh loops to one place, refreshLimits()
- added refreshWhen() which returns approximate time when
refreshCheck() may return true
rousskov [Wed, 22 Apr 1998 22:21:58 +0000 (22:21 +0000)]
- use store_status == STORE_OK to check if we re-used old entry
- removed storeComplete() call -- http.c calls storeComplete()
- added digest cblock checks for bits per entry and other new fields
- disable digest if bit utilization is too high (peerDigestUseful)
rousskov [Wed, 22 Apr 1998 22:21:39 +0000 (22:21 +0000)]
- removed storeDigestScheduleRebuild() call
rousskov [Wed, 22 Apr 1998 22:21:24 +0000 (22:21 +0000)]
- CacheDigestVer upgraded to 2/2 because we record bits_per_entry
and other new essential fields; the cblock size changed as well
- renamed CacheDigestHashCount to CacheDigestHashFuncCount
rousskov [Wed, 22 Apr 1998 22:21:10 +0000 (22:21 +0000)]
- enabled 304 replies for ENTRY_SPECIAL
this will lead to coredumps on caches running CD version 1
- temporary support for X-Cache-Lookup header field/stats
rousskov [Wed, 22 Apr 1998 22:20:55 +0000 (22:20 +0000)]
- configurable number of bits_per_entry per digest
- fixed resizing bug (did not set new bits to 0)
- removed public cacheDigestSizedCreate(),
simplified creation interface
- fixed cacheDigestDestroy bug (called xfree instead of memFree)
- temporary support for X-Cache-Lookup header field/stats
rousskov [Wed, 22 Apr 1998 11:54:28 +0000 (11:54 +0000)]
- temporary fix to prevent digests from overflows
wessels [Wed, 22 Apr 1998 11:41:11 +0000 (11:41 +0000)]
remove debug-openfd hack
wessels [Wed, 22 Apr 1998 08:07:32 +0000 (08:07 +0000)]
Changed error message to 'generated TIME by HOST (SQUID/VER)' because
anything with an '@' just MUST be an email address.
wessels [Wed, 22 Apr 1998 07:40:15 +0000 (07:40 +0000)]
initalize lru_list to NULL
wessels [Wed, 22 Apr 1998 04:49:15 +0000 (04:49 +0000)]
update
wessels [Wed, 22 Apr 1998 04:43:44 +0000 (04:43 +0000)]
re-fix parse_rfc1123() DT/ST bugs. Use the parsed time to generate
the GMT offset. DONT use the current time, because it will be different
for DT and ST.
wessels [Wed, 22 Apr 1998 03:33:20 +0000 (03:33 +0000)]
update aclCheckFast(no_cache) comments
rousskov [Wed, 22 Apr 1998 03:28:13 +0000 (03:28 +0000)]
- changed semantics of "no_cache" config tag from "allow" to "deny"
and synced the corresponding if statement in clientCachable()
rousskov [Wed, 22 Apr 1998 03:24:07 +0000 (03:24 +0000)]
- changed semantics of "no_cache" config tag from "allow" to "deny"
wessels [Wed, 22 Apr 1998 02:49:29 +0000 (02:49 +0000)]
update
wessels [Wed, 22 Apr 1998 02:49:05 +0000 (02:49 +0000)]
Added full FTP failure message. I'm sure someone will bitch about
this HTML though
wessels [Wed, 22 Apr 1998 02:41:20 +0000 (02:41 +0000)]
Fixed ftpParseListing loop
Added full error message to ERR_FTP_FAIL error pages
wessels [Tue, 21 Apr 1998 22:40:50 +0000 (22:40 +0000)]
Move 'fake a recent reply' hack to before neighborUp() check.
Joe Ramey
wessels [Tue, 21 Apr 1998 22:39:00 +0000 (22:39 +0000)]
Fixed hierarchy_stoplist to be DEFAULT_IF_NONE
wessels [Tue, 21 Apr 1998 08:20:18 +0000 (08:20 +0000)]
update
wessels [Tue, 21 Apr 1998 08:20:07 +0000 (08:20 +0000)]
in storeMaintainSwapSpace, move locked objects from the
tail to the head of the LRU list.
wessels [Tue, 21 Apr 1998 08:10:26 +0000 (08:10 +0000)]
these client-side-ish things should not be calling storeComplete
wessels [Tue, 21 Apr 1998 06:29:29 +0000 (06:29 +0000)]
fix fix
wessels [Tue, 21 Apr 1998 05:26:12 +0000 (05:26 +0000)]
Added storeSwapOutWriteQueued().
storeAbort() was improperly checking for no-writes-queued condition
by not accounting for the swap meta header size.
wessels [Tue, 21 Apr 1998 03:41:22 +0000 (03:41 +0000)]
open swapout fd debugging
wessels [Tue, 21 Apr 1998 00:32:46 +0000 (00:32 +0000)]
got acl match logic wrong
rousskov [Sat, 18 Apr 1998 12:54:08 +0000 (12:54 +0000)]
- fixed bit sequence calculation bug
- added #ifdef arround cd. histogram update
rousskov [Sat, 18 Apr 1998 11:41:54 +0000 (11:41 +0000)]
- removed call to peerDigestLookup from peerHandleIcpReply when peer
(or request) is NULL
wessels [Fri, 17 Apr 1998 10:25:35 +0000 (10:25 +0000)]
coredump fix. we called clientSendMoreData with an error indication
from storeUnregister. http->entry == NULL, but we assumed it could
not be. Similar assumption in clientWriteComplete.
wessels [Fri, 17 Apr 1998 10:24:25 +0000 (10:24 +0000)]
we are forgetting to reset filemap bits in storeRelease!
wessels [Fri, 17 Apr 1998 10:24:07 +0000 (10:24 +0000)]
we can use storeUnlinkFileno instead of safeunlink
wessels [Fri, 17 Apr 1998 03:51:54 +0000 (03:51 +0000)]
need to copy old filemap to new filemap during reconfigure
wessels [Fri, 17 Apr 1998 00:32:53 +0000 (00:32 +0000)]
fix SNMP typedefs
wessels [Fri, 17 Apr 1998 00:14:54 +0000 (00:14 +0000)]
make the filemap size dependent on the cache dir size and average
object size
wessels [Fri, 17 Apr 1998 00:14:33 +0000 (00:14 +0000)]
must use allow keyword for no_cache access list
wessels [Fri, 17 Apr 1998 00:06:32 +0000 (00:06 +0000)]
- Replaced cache_stoplist with 'no_cache' access list
- started coding SRC_RTT acl
- s/struct _// on structs.h;
wessels [Thu, 16 Apr 1998 23:48:50 +0000 (23:48 +0000)]
rename CACHE_DIGEST_MISS to NO_CACHE_DIGEST_DIRECT
wessels [Thu, 16 Apr 1998 23:47:27 +0000 (23:47 +0000)]
lock entries during abort
wessels [Thu, 16 Apr 1998 23:47:15 +0000 (23:47 +0000)]
show percent of file map bits used
wessels [Thu, 16 Apr 1998 23:18:52 +0000 (23:18 +0000)]
Add stat counter for # filemap bits in use
rousskov [Wed, 15 Apr 1998 06:34:34 +0000 (06:34 +0000)]
- added printouts of server_svc_time histograms
rousskov [Tue, 14 Apr 1998 22:38:21 +0000 (22:38 +0000)]
- added cd.on_xition_count histogram to Counter
counts the number of 0->1 transitions during cacheDigestAdd()
rousskov [Tue, 14 Apr 1998 21:16:23 +0000 (21:16 +0000)]
- digest size is now calculated using both maximum cache size
and currently used size:
the capacity is set to 1.5*#store_entries and then fitted
inside [10%,100%] of the max number of entries in the store hash
size = FudgeFactor * capacity;
- added debug to report who is requesting digests from us
current level is 1
we probably want to maintain a list (like the "bad peers sending IPC" list)
rather then reporting it via cache.log
- maybe-fix to determine whether we should call StoreComplete()
in peer_digest.c (reused old entryXor not?). Looks ugly though.X
rousskov [Mon, 13 Apr 1998 06:28:39 +0000 (06:28 +0000)]
(3) deny/allow tags for HTTP header fields (AR)
rousskov [Sun, 12 Apr 1998 12:13:56 +0000 (12:13 +0000)]
- temporary disabled to 304 replies for special entries to prevent coredumps
on caches running old buggy code that could not handle 304 replies for Cache
Digests
rousskov [Sun, 12 Apr 1998 12:10:05 +0000 (12:10 +0000)]
- client_side: fixed processing for IMS requests for special entries
the earlier code returned 200 replies instead of 304s
- peer_digest: bug fix: 304 replies caused coredumps
minor debug() polishing
- stat:
- structs:
- peer_select: added histograms for #choices when selecting a peer with
Cache Digest
- protos: added statHistIntDumper
rousskov [Sun, 12 Apr 1998 12:00:44 +0000 (12:00 +0000)]
- added dumper for integer-based histograms
rousskov [Sat, 11 Apr 1998 09:10:05 +0000 (09:10 +0000)]
(5) replace "for" loops in md5.c with memcpy and memset
rousskov [Sat, 11 Apr 1998 04:57:34 +0000 (04:57 +0000)]
- Bug fix: changed LOG_TAG_NOPNE to LOG_ICP_QUERY in all icpUdpSend() calls.
wessels [Fri, 10 Apr 1998 07:27:29 +0000 (07:27 +0000)]
updating wrong value for server_svc_time
wessels [Fri, 10 Apr 1998 07:13:05 +0000 (07:13 +0000)]
updating wrong value for server_svc_time
wessels [Fri, 10 Apr 1998 06:59:28 +0000 (06:59 +0000)]
update
wessels [Fri, 10 Apr 1998 06:51:04 +0000 (06:51 +0000)]
missed a new urlInternal
rousskov [Fri, 10 Apr 1998 06:49:39 +0000 (06:49 +0000)]
- polish
rousskov [Fri, 10 Apr 1998 06:45:22 +0000 (06:45 +0000)]
- polish