moparisthebest/curl
1
0
Fork 0
mirror of https://github.com/moparisthebest/curl synced 2024-12-22 08:08:50 -05:00
Code Issues Releases Wiki Activity

INTERNALS: absorbed docs/LIBCURL-STRUCTS

Browse Source
This commit is contained in:
Daniel Stenberg 2015-06-10 00:11:54 +02:00
parent 55f3eb588d
commit 1514f3506b
3 changed files with 234 additions and 250 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

234
docs/INTERNALS
View File

@ -37,6 +37,7 @@ Table of Contents
- [hostip.c explained](#hostip)
- [Track Down Memory Leaks](#memoryleak)
- [`multi_socket`](#multi_socket)
- [Structs in libcurl](#structs)
<a name="intro"></a>
curl internals
@ -848,6 +849,235 @@ Track Down Memory Leaks
only `fd_sets` and that is no longer good enough. The changes done to c-ares
are available in c-ares 1.3.1 and later.
<a name="structs"></a>
Structs in libcurl
==================
This section should cover 7.32.0 pretty accurately, but will make sense even
for older and later versions as things don't change drastically that often.
## SessionHandle
The SessionHandle handle struct is the one returned to the outside in the
external API as a "CURL *". This is usually known as an easy handle in API
documentations and examples.
Information and state that is related to the actual connection is in the
'connectdata' struct. When a transfer is about to be made, libcurl will
either create a new connection or re-use an existing one. The particular
connectdata that is used by this handle is pointed out by
SessionHandle->easy_conn.
Data and information that regard this particular single transfer is put in
the SingleRequest sub-struct.
When the SessionHandle struct is added to a multi handle, as it must be in
order to do any transfer, the ->multi member will point to the `Curl_multi`
struct it belongs to. The ->prev and ->next members will then be used by the
multi code to keep a linked list of SessionHandle structs that are added to
that same multi handle. libcurl always uses multi so ->multi *will* point to
a `Curl_multi` when a transfer is in progress.
->mstate is the multi state of this particular SessionHandle. When
`multi_runsingle()` is called, it will act on this handle according to which
state it is in. The mstate is also what tells which sockets to return for a
specific SessionHandle when [`curl_multi_fdset()`][12] is called etc.
The libcurl source code generally use the name 'data' for the variable that
points to the SessionHandle.
When doing multiplexed HTTP/2 transfers, each SessionHandle is associated
with an individual stream, sharing the same connectdata struct. Multiplexing
makes it even more important to keep things associated with the right thing!
## connectdata
A general idea in libcurl is to keep connections around in a connection
"cache" after they have been used in case they will be used again and then
re-use an existing one instead of creating a new as it creates a significant
performance boost.
Each 'connectdata' identifies a single physical connection to a server. If
the connection can't be kept alive, the connection will be closed after use
and then this struct can be removed from the cache and freed.
Thus, the same SessionHandle can be used multiple times and each time select
another connectdata struct to use for the connection. Keep this in mind, as
it is then important to consider if options or choices are based on the
connection or the SessionHandle.
Functions in libcurl will assume that connectdata->data points to the
SessionHandle that uses this connection (for the moment).
As a special complexity, some protocols supported by libcurl require a
special disconnect procedure that is more than just shutting down the
socket. It can involve sending one or more commands to the server before
doing so. Since connections are kept in the connection cache after use, the
original SessionHandle may no longer be around when the time comes to shut
down a particular connection. For this purpose, libcurl holds a special
dummy `closure_handle` SessionHandle in the `Curl_multi` struct to use when
needed.
FTP uses two TCP connections for a typical transfer but it keeps both in
this single struct and thus can be considered a single connection for most
internal concerns.
The libcurl source code generally use the name 'conn' for the variable that
points to the connectdata.
## Curl_multi
Internally, the easy interface is implemented as a wrapper around multi
interface functions. This makes everything multi interface.
`Curl_multi` is the multi handle struct exposed as "CURLM *" in external APIs.
This struct holds a list of SessionHandle structs that have been added to
this handle with [`curl_multi_add_handle()`][13]. The start of the list is
->easyp and ->num_easy is a counter of added SessionHandles.
->msglist is a linked list of messages to send back when
[`curl_multi_info_read()`][14] is called. Basically a node is added to that
list when an individual SessionHandle's transfer has completed.
->hostcache points to the name cache. It is a hash table for looking up name
to IP. The nodes have a limited life time in there and this cache is meant
to reduce the time for when the same name is wanted within a short period of
time.
->timetree points to a tree of SessionHandles, sorted by the remaining time
until it should be checked - normally some sort of timeout. Each
SessionHandle has one node in the tree.
->sockhash is a hash table to allow fast lookups of socket descriptor to
which SessionHandle that uses that descriptor. This is necessary for the
`multi_socket` API.
->conn_cache points to the connection cache. It keeps track of all
connections that are kept after use. The cache has a maximum size.
->closure_handle is described in the 'connectdata' section.
The libcurl source code generally use the name 'multi' for the variable that
points to the Curl_multi struct.
## Curl_handler
Each unique protocol that is supported by libcurl needs to provide at least
one `Curl_handler` struct. It defines what the protocol is called and what
functions the main code should call to deal with protocol specific issues.
In general, there's a source file named [protocol].c in which there's a
"struct `Curl_handler` `Curl_handler_[protocol]`" declared. In url.c there's
then the main array with all individual `Curl_handler` structs pointed to
from a single array which is scanned through when a URL is given to libcurl
to work with.
->scheme is the URL scheme name, usually spelled out in uppercase. That's
"HTTP" or "FTP" etc. SSL versions of the protcol need its own `Curl_handler`
setup so HTTPS separate from HTTP.
->setup_connection is called to allow the protocol code to allocate protocol
specific data that then gets associated with that SessionHandle for the rest
of this transfer. It gets freed again at the end of the transfer. It will be
called before the 'connectdata' for the transfer has been selected/created.
Most protocols will allocate its private 'struct [PROTOCOL]' here and assign
SessionHandle->req.protop to point to it.
->connect_it allows a protocol to do some specific actions after the TCP
connect is done, that can still be considered part of the connection phase.
Some protocols will alter the connectdata->recv[] and connectdata->send[]
function pointers in this function.
->connecting is similarly a function that keeps getting called as long as the
protocol considers itself still in the connecting phase.
->do_it is the function called to issue the transfer request. What we call
the DO action internally. If the DO is not enough and things need to be kept
getting done for the entire DO sequence to complete, ->doing is then usually
also provided. Each protocol that needs to do multiple commands or similar
for do/doing need to implement their own state machines (see SCP, SFTP,
FTP). Some protocols (only FTP and only due to historical reasons) has a
separate piece of the DO state called `DO_MORE`.
->doing keeps getting called while issuing the transfer request command(s)
->done gets called when the transfer is complete and DONE. That's after the
main data has been transferred.
->do_more gets called during the `DO_MORE` state. The FTP protocol uses this
state when setting up the second connection.
->`proto_getsock`
->`doing_getsock`
->`domore_getsock`
->`perform_getsock`
Functions that return socket information. Which socket(s) to wait for which
action(s) during the particular multi state.
->disconnect is called immediately before the TCP connection is shutdown.
->readwrite gets called during transfer to allow the protocol to do extra
reads/writes
->defport is the default report TCP or UDP port this protocol uses
->protocol is one or more bits in the `CURLPROTO_*` set. The SSL versions
have their "base" protocol set and then the SSL variation. Like
"HTTP|HTTPS".
->flags is a bitmask with additional information about the protocol that will
make it get treated differently by the generic engine:
- `PROTOPT_SSL` - will make it connect and negotiate SSL
- `PROTOPT_DUAL` - this protocol uses two connections
- `PROTOPT_CLOSEACTION` - this protocol has actions to do before closing the
connection. This flag is no longer used by code, yet still set for a bunch
protocol handlers.
- `PROTOPT_DIRLOCK` - "direction lock". The SSH protocols set this bit to
limit which "direction" of socket actions that the main engine will
concern itself about.
- `PROTOPT_NONETWORK` - a protocol that doesn't use network (read file:)
- `PROTOPT_NEEDSPWD` - this protocol needs a password and will use a default
one unless one is provided
- `PROTOPT_NOURLQUERY` - this protocol can't handle a query part on the URL
(?foo=bar)
## conncache
Is a hash table with connections for later re-use. Each SessionHandle has
a pointer to its connection cache. Each multi handle sets up a connection
cache that all added SessionHandles share by default.
## Curl_share
The libcurl share API allocates a `Curl_share` struct, exposed to the
external API as "CURLSH *".
The idea is that the struct can have a set of own versions of caches and
pools and then by providing this struct in the `CURLOPT_SHARE` option, those
specific SessionHandles will use the caches/pools that this share handle
holds.
Then individual SessionHandle structs can be made to share specific things
that they otherwise wouldn't, such as cookies.
The `Curl_share` struct can currently hold cookies, DNS cache and the SSL
session cache.
## CookieInfo
This is the main cookie struct. It holds all known cookies and related
information. Each SessionHandle has its own private CookieInfo even when
they are added to a multi handle. They can be made to share cookies by using
the share API.
[1]: http://curl.haxx.se/libcurl/c/curl_easy_setopt.html
[2]: http://curl.haxx.se/libcurl/c/curl_easy_init.html
@ -860,4 +1090,6 @@ Track Down Memory Leaks
[9]: http://curl.haxx.se/libcurl/c/curl_multi_setopt.html
[10]: http://curl.haxx.se/libcurl/c/CURLMOPT_TIMERFUNCTION.html
[11]: http://curl.haxx.se/libcurl/c/curl_multi_perform.html
[12]: http://curl.haxx.se/libcurl/c/curl_multi_fdset.html
[12]: http://curl.haxx.se/libcurl/c/curl_multi_fdset.html
[13]: http://curl.haxx.se/libcurl/c/curl_multi_add_handle.html
[14]: http://curl.haxx.se/libcurl/c/curl_multi_info_read.html

248
docs/LIBCURL-STRUCTS
View File

@ -1,248 +0,0 @@
_ _ ____ _
___| | | | _ \| |
/ __| | | | |_) | |
| (__| |_| | _ <| |___
\___|\___/|_| \_\_____|
Structs in libcurl
This document should cover 7.32.0 pretty accurately, but will make sense even
for older and later versions as things don't change drastically that often.
1. The main structs in libcurl
1.1 SessionHandle
1.2 connectdata
1.3 Curl_multi
1.4 Curl_handler
1.5 conncache
1.6 Curl_share
1.7 CookieInfo
==============================================================================
1. The main structs in libcurl
1.1 SessionHandle
The SessionHandle handle struct is the one returned to the outside in the
external API as a "CURL *". This is usually known as an easy handle in API
documentations and examples.
Information and state that is related to the actual connection is in the
'connectdata' struct. When a transfer is about to be made, libcurl will
either create a new connection or re-use an existing one. The particular
connectdata that is used by this handle is pointed out by
SessionHandle->easy_conn.
Data and information that regard this particular single transfer is put in
the SingleRequest sub-struct.
When the SessionHandle struct is added to a multi handle, as it must be in
order to do any transfer, the ->multi member will point to the Curl_multi
struct it belongs to. The ->prev and ->next members will then be used by the
multi code to keep a linked list of SessionHandle structs that are added to
that same multi handle. libcurl always uses multi so ->multi *will* point to
a Curl_multi when a transfer is in progress.
->mstate is the multi state of this particular SessionHandle. When
multi_runsingle() is called, it will act on this handle according to which
state it is in. The mstate is also what tells which sockets to return for a
specific SessionHandle when curl_multi_fdset() is called etc.
The libcurl source code generally use the name 'data' for the variable that
points to the SessionHandle.
When doing multiplexed HTTP/2 transfers, each SessionHandle is associated
with an individual stream, sharing the same connectdata struct. Multiplexing
makes it even more important to keep things associated with the right thing!
1.2 connectdata
A general idea in libcurl is to keep connections around in a connection
"cache" after they have been used in case they will be used again and then
re-use an existing one instead of creating a new as it creates a significant
performance boost.
Each 'connectdata' identifies a single physical connection to a server. If
the connection can't be kept alive, the connection will be closed after use
and then this struct can be removed from the cache and freed.
Thus, the same SessionHandle can be used multiple times and each time select
another connectdata struct to use for the connection. Keep this in mind, as
it is then important to consider if options or choices are based on the
connection or the SessionHandle.
Functions in libcurl will assume that connectdata->data points to the
SessionHandle that uses this connection (for the moment).
As a special complexity, some protocols supported by libcurl require a
special disconnect procedure that is more than just shutting down the
socket. It can involve sending one or more commands to the server before
doing so. Since connections are kept in the connection cache after use, the
original SessionHandle may no longer be around when the time comes to shut
down a particular connection. For this purpose, libcurl holds a special
dummy 'closure_handle' SessionHandle in the Curl_multi struct to
FTP uses two TCP connections for a typical transfer but it keeps both in
this single struct and thus can be considered a single connection for most
internal concerns.
The libcurl source code generally use the name 'conn' for the variable that
points to the connectdata.
1.3 Curl_multi
Internally, the easy interface is implemented as a wrapper around multi
interface functions. This makes everything multi interface.
Curl_multi is the multi handle struct exposed as "CURLM *" in external APIs.
This struct holds a list of SessionHandle structs that have been added to
this handle with curl_multi_add_handle(). The start of the list is ->easyp
and ->num_easy is a counter of added SessionHandles.
->msglist is a linked list of messages to send back when
curl_multi_info_read() is called. Basically a node is added to that list
when an individual SessionHandle's transfer has completed.
->hostcache points to the name cache. It is a hash table for looking up name
to IP. The nodes have a limited life time in there and this cache is meant
to reduce the time for when the same name is wanted within a short period of
time.
->timetree points to a tree of SessionHandles, sorted by the remaining time
until it should be checked - normally some sort of timeout. Each
SessionHandle has one node in the tree.
->sockhash is a hash table to allow fast lookups of socket descriptor to
which SessionHandle that uses that descriptor. This is necessary for the
multi_socket API.
->conn_cache points to the connection cache. It keeps track of all
connections that are kept after use. The cache has a maximum size.
->closure_handle is described in the 'connectdata' section.
The libcurl source code generally use the name 'multi' for the variable that
points to the Curl_multi struct.
1.4 Curl_handler
Each unique protocol that is supported by libcurl needs to provide at least
one Curl_handler struct. It defines what the protocol is called and what
functions the main code should call to deal with protocol specific issues.
In general, there's a source file named [protocol].c in which there's a
"struct Curl_handler Curl_handler_[protocol]" declared. In url.c there's
then the main array with all individual Curl_handler structs pointed to from
a single array which is scanned through when a URL is given to libcurl to
work with.
->scheme is the URL scheme name, usually spelled out in uppercase. That's
"HTTP" or "FTP" etc. SSL versions of the protcol need its own Curl_handler
setup so HTTPS separate from HTTP.
->setup_connection is called to allow the protocol code to allocate protocol
specific data that then gets associated with that SessionHandle for the rest
of this transfer. It gets freed again at the end of the transfer. It will be
called before the 'connectdata' for the transfer has been selected/created.
Most protocols will allocate its private 'struct [PROTOCOL]' here and assign
SessionHandle->req.protop to point to it.
->connect_it allows a protocol to do some specific actions after the TCP
connect is done, that can still be considered part of the connection phase.
Some protocols will alter the connectdata->recv[] and connectdata->send[]
function pointers in this function.
->connecting is similarly a function that keeps getting called as long as the
protocol considers itself still in the connecting phase.
->do_it is the function called to issue the transfer request. What we call
the DO action internally. If the DO is not enough and things need to be kept
getting done for the entire DO sequence to complete, ->doing is then usually
also provided. Each protocol that needs to do multiple commands or similar
for do/doing need to implement their own state machines (see SCP, SFTP,
FTP). Some protocols (only FTP and only due to historical reasons) has a
separate piece of the DO state called DO_MORE.
->doing keeps getting called while issuing the transfer request command(s)
->done gets called when the transfer is complete and DONE. That's after the
main data has been transferred.
->do_more gets called during the DO_MORE state. The FTP protocol uses this
state when setting up the second connection.
->proto_getsock
->doing_getsock
->domore_getsock
->perform_getsock
Functions that return socket information. Which socket(s) to wait for which
action(s) during the particular multi state.
->disconnect is called immediately before the TCP connection is shutdown.
->readwrite gets called during transfer to allow the protocol to do extra
reads/writes
->defport is the default report TCP or UDP port this protocol uses
->protocol is one or more bits in the CURLPROTO_* set. The SSL versions have
their "base" protocol set and then the SSL variation. Like "HTTP|HTTPS".
->flags is a bitmask with additional information about the protocol that will
make it get treated differently by the generic engine:
PROTOPT_SSL - will make it connect and negotiate SSL
PROTOPT_DUAL - this protocol uses two connections
PROTOPT_CLOSEACTION - this protocol has actions to do before closing the
connection. This flag is no longer used by code, yet still set for a bunch
protocol handlers.
PROTOPT_DIRLOCK - "direction lock". The SSH protocols set this bit to
limit which "direction" of socket actions that the main engine will
concern itself about.
PROTOPT_NONETWORK - a protocol that doesn't use network (read file:)
PROTOPT_NEEDSPWD - this protocol needs a password and will use a default
one unless one is provided
PROTOPT_NOURLQUERY - this protocol can't handle a query part on the URL
(?foo=bar)
1.5 conncache
Is a hash table with connections for later re-use. Each SessionHandle has
a pointer to its connection cache. Each multi handle sets up a connection
cache that all added SessionHandles share by default.
1.6 Curl_share
The libcurl share API allocates a Curl_share struct, exposed to the external
API as "CURLSH *".
The idea is that the struct can have a set of own versions of caches and
pools and then by providing this struct in the CURLOPT_SHARE option, those
specific SessionHandles will use the caches/pools that this share handle
holds.
Then individual SessionHandle structs can be made to share specific things
that they otherwise wouldn't, such as cookies.
The Curl_share struct can currently hold cookies, DNS cache and the SSL
session cache.
1.7 CookieInfo
This is the main cookie struct. It holds all known cookies and related
information. Each SessionHandle has its own private CookieInfo even when
they are added to a multi handle. They can be made to share cookies by using
the share API.

2
docs/Makefile.am
View File

@ -37,7 +37,7 @@ EXTRA_DIST = MANUAL BUGS CONTRIBUTE FAQ FEATURES INTERNALS SSLCERTS \
README.win32 RESOURCES TODO TheArtOfHttpScripting THANKS VERSIONS \
KNOWN_BUGS BINDINGS $(man_MANS) $(HTMLPAGES) HISTORY INSTALL \
$(PDFPAGES) LICENSE-MIXING README.netware DISTRO-DILEMMA INSTALL.devcpp \
MAIL-ETIQUETTE HTTP-COOKIES LIBCURL-STRUCTS SECURITY RELEASE-PROCEDURE \
MAIL-ETIQUETTE HTTP-COOKIES SECURITY RELEASE-PROCEDURE \
SSL-PROBLEMS HTTP2.md ROADMAP.md
MAN2HTML= roffit < $< >$@