The following commits introduced performance improvements and bugfixes:
Author: Xavier Del Campo Romero <xavi.dcr@tutanota.com>
Date: Sat Jan 20 01:09:18 2024 +0100
server.c: Fix wrong priority for do_exit
Under some specific circumstances, poll(2) would return a positive
integer, but do_exit might had been previously set. This caused libweb
to ignore SIGTERM, with the potential risk for an endless loop.
Author: Xavier Del Campo Romero <xavi.dcr@tutanota.com>
Date: Sat Jan 20 01:05:05 2024 +0100
http.c: Solve performance issues on POST uploads
Profiling showed that reading multipart/form POST uploads byte-by-byte
was too slow and typically led to maximum CPU usage. Therefore, the
older approach (as done up to commit 7efc2b3a) was more efficient, even
if the resulting code was a bit uglier.
slweb was renamed to libweb some time ago:
commit 28ae865e5e
Author: Xavier Del Campo Romero <xavi.dcr@tutanota.com>
Date: Tue Oct 10 23:43:47 2023 +0200
Apply slweb renaming to libweb
The following commit should increase performance for slcl:
commit b0accd099fa8c5110d4c3c68830ad6fd810ca3ec
Author: Xavier Del Campo Romero <xavi.dcr@tutanota.com>
Date: Fri Nov 24 00:52:50 2023 +0100
http.c: Unify read operations
So far, libweb would perform different read operations depending on its
state:
- For HTTP headers or request bodies, one byte at a time was read.
- For multipart/form-data, up to BUFSIZ bytes at a time were read.
However, this caused a significant extra number of syscalls for no
reason and would increase code complexity, specially when parsing
multiform/form-data boundaries.
Now, http_read always reads up to BUFSIZ bytes at a time and process
them on a loop. Apart from reducing code complexity, this should
increase performance due to the (much) lower number of syscalls
required.
Recent commits from libweb brought a few breaking changes. The one below
affected slcl, so it had to be updated according to the new interface:
commit 98f5f52461b0c1ab1ee3331722bd32e2db9e1d41
Author: Xavier Del Campo <xavier.delcampo@midokura.com>
Date: Thu Nov 16 12:23:08 2023 +0100
Split handler_loop from handler_listen
Some applications might set up a struct handler object to listen on any
port i.e., 0, but still need a way to determine which port number was
eventually selected by the implementation.
Therefore, handler_listen has been reduced to the server initialization
bit, whereas the main loop has been split into its own function, namely
handler_loop.
Because of these changes, it no longer made sense for libweb to write
the selected port to standard output, as this is something now
applications can do on their own.
When a user attempts to upload a file into a non-existing directory,
slcl would not check whether the directory exists. Then, rename(3) would
fail and slcl would treat this as a fatal error, effectively closing
itself.
Since this is an example of ill-formed user input, it must be treated as
a non-fatal error, and instead slcl should return a bad request page.
More bugfixes related to partial boundary parsing were provided by this
commit:
commit b71a6174e12b4709acaf8bc151938ba12d2a54f6
Author: Xavier Del Campo Romero <xavi.dcr@tutanota.com>
Date: Sun Nov 12 23:31:57 2023 +0100
http.c: Fix more issues with partial boundaries
- http_memmem must not check strlen(a) > n because, in case of a partial
boundary, it would wrongfully return NULL.
- If one or more characters from a partial boundary are found at the end
of a buffer, but the next buffer does not start with the rest of the
boundary, the accumulated boundary must be reset, and then look for a
new boundary.
Several bugfixes related to partial boundary parsing were provided by
this commit:
commit 7d02b225fe11fb0c7233cd2ea576485ee920f203
Author: Xavier Del Campo Romero <xavi.dcr@tutanota.com>
Date: Sun Nov 12 06:16:26 2023 +0100
http.c: Fix several issues with partial boundaries
- Writing to m->boundary[len] did not make any sense, as len is not
meant to change between calls to read_mf_boundary_byte.
- For the same reason, memset(3)ing "len + 1" did not make any sense.
- When a partial boundary is found, http_memmem must still return st.
- Calling reset_boundary with prev == 0 did not make sense, since that
case typically means a partial boundary was found on a previous
iteration, so m->blen must not be reset.
The new signature allows callers to distinguish decoding errors from
fatal errors. This is important for slcl to avoid crashing when
ill-formed data is received from a client.
Despite common use in several POSIX operating systems, mktemp(1) is not
defined by POSIX.1-2008, nor even POSIX.1-2017. As long as it is not
introduced, m4(1)'s mkstemp can be used with similar effect.
According to C99 7.19.1p3:
BUFSIZ is a macro that expands to an integer constant expression that is
the size of the buffer used by the setbuf function.
In other words, this means BUFSIZ is the most optimal length for a
buffer that reads a file into memory in chunks using fread(3).
For platforms where int is a 16-bit data type, this operation might
overflow and possibly cause either unexpected behaviour and/or a
compiler warning.
Therefore, it is safer to promote each integer constant accordingly.
This feature was already implemented by:
commit 0822a982ef
Author: Xavier Del Campo Romero <xavi.dcr@tutanota.com>
Date: Sat Jul 8 00:54:59 2023 +0200
Implement file/directory removal
sha256sum(1) is a GNU utility that might not be available under some
POSIX systems. Since OpenSSL is already a dependency, it makes sense to
reuse it to generate SHA256 digests.
slweb now assumes application/x-www-form-urlencoded-data as text, so it
now returns a null-terminated string on struct http_post member "data".
This removes the need for slcl to call strdup(3) in order to obtain a
null-terminated string.
For historical reasons, slweb used to check for a name called "dir" on
multipart/form-data POST requests. However, stricly speaking this is
application logic, so it has been now moved from slweb to slcl.
This has resulted in a couple of breaking changes in slweb that had to
be updated on slcl.
d55b84f68b90fe5c2521724d0c22ebf22e62b0b4 introduces a bugfix:
Author: Xavier Del Campo Romero <xavi.dcr@tutanota.com>
Date: Tue Aug 8 00:32:21 2023 +0200
html.c: Fix wrong encoding for '>' and '<'
