rdelaage
/
gmid
mirror of https://github.com/omar-polo/gmid.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
408 Commits 2 Branches 31 Tags 6.5 MiB
Commit Graph

85 Commits

Author SHA1 Message Date
Omar Polo 62e001b067 move all sandbox-related code to sandbox.c 
while there, add capsicum for the logger process
 2021-03-20 08:42:08 +00:00
Omar Polo bc99d868bc refactoring: imsg everywhere 
use imsg to handle ALL kinds of IPC in gmid.  This simplifies and shorten the
code, and  makes everything more uniform too.
 2021-03-19 19:21:29 +00:00
Omar Polo 2c3e53dac6 give each server process its own socket for the executor 
this fixes a bug introduced with the prefork mechanics: every server
process shared the same socket, and this would cause a race condition
when multiple server processes asked for a script cgi being executed.

This gives each server process its own socket to talk to the executor,
so the race cannot happen.
 2021-03-03 17:22:01 +00:00
Omar Polo 376a540764
move log_init & vars to gmid.c, retain logger_main in log.c 
this is to let the regression suite compile
 2021-02-23 13:44:20 +01:00
Omar Polo 793835cb26
add `log on/off' to enable/disable logs per-location 2021-02-23 13:43:24 +01:00
Omar Polo d278a0c3c5
moving logging to its own process 2021-02-23 13:40:59 +01:00
Omar Polo 52418c8d82 fix various compilation errors 
Include gmid.h as first header in every file, as it then includes
config.h (that defines _GNU_SOURCE for instance).

Fix also a warning about unsigned vs signed const char pointers in
openssl.
 2021-02-12 12:47:20 +00:00
Omar Polo 9356f61a63 BUFSIZ is variable, we need *exactly* 1024 
Using BUFSIZ in sbuf is not OK.  It's variable, and in various places
we assume that sbuf is 1024 (like handle_cgi_reply).  We could patch
those, but we aren't sure BUFSIZ is >= 1024!  Let's keep the hardcoded
number.

(found by debugging on arch on amd64, where BUFSIZ is bigger)
 2021-02-12 11:56:13 +00:00
Omar Polo 27b2fa9ae5 don't mmap 
Before we mmap(2) file for reading, and use a buffer to handle CGI
scripts.  Turns out, for sequential access over the whole mmap isn't
better than our loop on read.  This has also the additional advantage
that we can use handle_cgi (now handle_copy) for both files and CGI,
which is pretty cool.

This also fixes a nasty bug where we could hang a connection forever,
because we scheduled the wrong type of event (read on POLLOUT and
write on POLLIN, it's the other way around!)
 2021-02-12 11:27:33 +00:00
Omar Polo 02be96c6dd add `require client ca' rule to require certs signed by a CA 2021-02-09 22:30:04 +00:00
Omar Polo d090dc8491 define config_path as global variable 2021-02-08 12:44:34 +00:00
Omar Polo abc007d2b3 rewrite main loop using libevent 2021-02-08 10:01:45 +00:00
Omar Polo b63e30ff44 define TLS_CLIENT_NOT_BEFORE/NOT_AFTER in CGI scripts 2021-02-07 21:47:01 +00:00
Omar Polo 9f006a2127 [cgi] split the query in words if needed and add them to the argv 2021-02-07 18:55:04 +00:00
Omar Polo 3abf91b0b4 improve logs management 2021-02-07 15:30:28 +00:00
Omar Polo a709ddf5eb added prefork option 2021-02-07 12:05:32 +00:00
Omar Polo e3ddf39095 add the ``entrypoint'' option 2021-02-06 18:28:43 +00:00
Omar Polo 6abda252e9 added ``block return'' and ``strip'' options 2021-02-06 17:22:37 +00:00
Omar Polo ca21e10043 reload configuration on SIGHUP 2021-02-04 13:23:15 +00:00
Omar Polo 4e2e2ab1d3 refactor executor_main 
now it's symmetrical to listener_main().
 2021-02-03 16:37:53 +00:00
Omar Polo d672b8fba1 refactoring startup logic 2021-02-03 16:28:00 +00:00
Omar Polo 9b8f5ed2c0 revert commit 346f28eeaa 
keep mark_nonblock in utils.c, as otherwise the build for the regress
suite will fail (mark_nonblock needs fatal which is in gmid.c, and
we can't link gmid.o with the regress suite...)
 2021-02-03 14:16:39 +00:00
Omar Polo 346f28eeaa move mark_nonblock to utils.c 2021-02-02 23:03:33 +00:00
Omar Polo fe40638928 mark various functions as static 
By marking all those function as static, the compiler is free to do
more optimizations.  In addition, those functions are not used outside
server.c
 2021-02-02 23:01:09 +00:00
Omar Polo 5f715ce43f print the header in the directory listing 2021-02-02 09:48:32 +00:00
Omar Polo 35744950aa simplify handle_cgi 
Now that I got rid of the enum+switch, adding more state is easier.
Before, we used an hack to remember if we had read the CGI reply or
not (c->code = -1).

This introduces a new state, handle_cgi_reply that reads the CGI
script reply, logs it, and only then switches to handle_cgi.
handle_cgi itself is cleaner, now it only reads into c->sbuf and send
what it had red.

We even get, almost for free, the 42 error.  If read exists with -1 or
0 from in handle_cgi_reply, we return a proper error to the client.
We can extend this further in the future and also try to validate the
CGI reply (for now we're only looking for a \n).
 2021-02-01 22:04:51 +00:00
Omar Polo 92da82858b document the DFA 2021-02-01 20:24:14 +00:00
Omar Polo 112802ea31 client state machine: function pointers instead of enum+switch 2021-02-01 20:00:33 +00:00
Omar Polo 2fafa2d23e bring the CGI implementation in par with GLV-1.12556 2021-02-01 11:11:43 +00:00
Omar Polo 6016a593a3 invert the location precedence: first match wins 
It's how httpd(8) does it, and it allows us to call fnmatch less time
 2021-01-30 12:04:20 +00:00
Omar Polo a2fd801327 puny_decode: set an error string 2021-01-29 17:11:03 +00:00
Omar Polo 35cf19e3f3 punycode: refactoring 
use the correct term "label" instead of "component" when speaking
about the parts of a hostname.  Also, fix the sizes for max hostname
and label.
 2021-01-28 16:25:59 +00:00
Omar Polo 473e4531d5 delete unused constant 2021-01-28 16:24:48 +00:00
Omar Polo ef0f2ae88c remove unused function 2021-01-28 14:21:57 +00:00
Omar Polo 13ed2fb63f improve configuration parsing error 2021-01-27 16:45:45 +00:00
Omar Polo fe5967cd02 const-ify strings in struct location 2021-01-27 15:53:30 +00:00
Omar Polo 44ee1bac8b use starts_with in puny.c 2021-01-27 15:35:09 +00:00
Omar Polo c4f682f855 trim_req_iri: set error string 2021-01-27 15:05:16 +00:00
Omar Polo 46af8c6cfe drop the daemon config in favour of the -f flag 
Now it daemonize by default when running with a config, unless the -f
flag is given.
 2021-01-27 12:04:37 +00:00
Omar Polo 7957cbd9aa const-ify puny_decode (and add puny.c) 2021-01-27 11:21:23 +00:00
Omar Polo 3300cbe06a initial punycode support 2021-01-27 10:47:49 +00:00
Omar Polo 8443bff77a rework the configless mode: change flags and generate certs 2021-01-25 14:08:31 +00:00
Omar Polo ae08ec7da5 chroot & drop privileges 2021-01-25 10:30:07 +00:00
Omar Polo 252908e6bb added support for location blocks 2021-01-24 18:53:26 +00:00
Omar Polo c8b7433918 added support for location blocks 2021-01-24 14:11:40 +00:00
Omar Polo 07b0a14218 void-ify some functions 
their return value is no longer used, it's only confusing at this point.
 2021-01-24 09:54:44 +00:00
Omar Polo a87f662565 refactoring state management 
instead of having a flag to discern between two different behaviours
in S_SENDING, split that state into S_SENDING_FILE and S_SENDING_CGI
(this will also make it easier in the future to add other sending
states).  While there, also get rid of `goodbye' and make start_reply
advance the state machine by itself.
 2021-01-24 09:49:09 +00:00
Omar Polo e7a2a99b5a added index option 2021-01-24 09:14:01 +00:00
Omar Polo f890c8c54d use a helper to handle no-body replies 2021-01-22 13:58:54 +00:00
Omar Polo 36162ed86d rename goodbye -> close_conn 2021-01-22 13:45:27 +00:00