summary refs log tree commit diff
Commit message (Collapse)AuthorAge
* vsockclient: allow null application HEAD masterAlyssa Ross2021-03-24
| | | | | | | | | | | | | | | | | | | | | | I found myself running vsockclient 2 1024 true in a situation where it just mattered that a connection was made, and no data needed to be exchanged of it. That I found myself doing this is quite a strong indicator that the program to exec into should be optional. UCSPI-1996[1] doesn't specify what should happen if no application is provided, so I think this is a reasonable interpretation. I'm not sure if I feel the same for vsockserver(d), so I haven't made the corresponding change there. I'll wait to see if there's ever a need for it. [1]: https://cr.yp.to/proto/ucspi.txt Message-Id: <20210324182433.19110-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Remove unused includesAlyssa Ross2021-03-24
| | | | | Message-Id: <20210324135022.15874-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* configure: reject invalid argumentsAlyssa Ross2021-03-22
| | | | | | | | | | | | | | But we preserve the autoconf behaviour of only warning on invalid options. $ ./configure --test Warning: unrecognized options: --test $ ./configure test Error: unexpected argument: test Message-Id: <20210321193951.21671-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Set up clang-formatAlyssa Ross2021-03-22
| | | | | Message-ID: <20210321193902.21087-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsockserver: fix uninitialized variableAlyssa Ross2021-03-21
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Identified by clang-tidy: vsockserver.c:105:17: warning: 3rd function call argument is an uninitialized value [clang-analyzer-core.CallAndMessage] while ((opt = argz_next(binder_opts, binder_opts_len, opt))) ^ vsockserver.c:44:3: note: Assuming the condition is false argz_add(&binder_opts, &binder_opts_len, BINDIR "/vsockserver-socketbinder") || ^ vsockserver.c:44:3: note: Left side of '||' is false vsockserver.c:47:9: note: Assuming the condition is false while ((opt = getopt(argc, argv, "+1qQv")) != -1) { ^ vsockserver.c:47:2: note: Loop condition is false. Execution continues on line 68 while ((opt = getopt(argc, argv, "+1qQv")) != -1) { ^ vsockserver.c:68:6: note: Assuming 'alloc_failed' is false if (alloc_failed) ^ vsockserver.c:68:2: note: Taking false branch if (alloc_failed) ^ vsockserver.c:77:6: note: Assuming the condition is false if (optind > argc - 3) ^ vsockserver.c:77:2: note: Taking false branch if (optind > argc - 3) ^ vsockserver.c:81:6: note: Assuming the condition is false if (argz_add(&binder_opts, &binder_opts_len, "--") || ^ vsockserver.c:81:6: note: Left side of '||' is false vsockserver.c:82:6: note: Assuming the condition is false argz_add(&binder_opts, &binder_opts_len, argv[optind++]) || ^ vsockserver.c:81:6: note: Left side of '||' is false if (argz_add(&binder_opts, &binder_opts_len, "--") || ^ vsockserver.c:83:6: note: Assuming the condition is false argz_add(&binder_opts, &binder_opts_len, argv[optind++])) ^ vsockserver.c:81:2: note: Taking false branch if (argz_add(&binder_opts, &binder_opts_len, "--") || ^ vsockserver.c:89:6: note: Assuming the condition is false if (argz_append(&binder_opts, &binder_opts_len, daemon_opts, daemon_opts_len)) ^ vsockserver.c:89:2: note: Taking false branch if (argz_append(&binder_opts, &binder_opts_len, daemon_opts, daemon_opts_len)) ^ vsockserver.c:96:23: note: Assuming 'i' is >= 'argc' for (int i = optind; i < argc; i++) ^ vsockserver.c:96:2: note: Loop condition is false. Execution continues on line 100 for (int i = optind; i < argc; i++) ^ vsockserver.c:100:6: note: Assuming 'verbosity' is equal to all if (verbosity == all) { ^ vsockserver.c:100:2: note: Taking true branch if (verbosity == all) { ^ vsockserver.c:101:3: note: 'opt' declared without an initial value char *opt; ^ vsockserver.c:105:17: note: 3rd function call argument is an uninitialized value while ((opt = argz_next(binder_opts, binder_opts_len, opt))) Message-Id: <20210321145440.28563-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* exec: free argv if exec failsAlyssa Ross2021-03-21
| | | | | | | | | | | | | | | | | | | | | | Identified by clang-tidy: exec.c:21:2: warning: Potential leak of memory pointed to by 'argv' [clang-analyzer-unix.Malloc] return execvp(file, argv); ^ exec.c:15:16: note: Memory is allocated char **argv = calloc(argz_count(argz, len) + 1, sizeof(char *)); ^ exec.c:16:6: note: Assuming 'argv' is non-null if (!argv) ^ exec.c:16:2: note: Taking false branch if (!argv) ^ exec.c:21:2: note: Potential leak of memory pointed to by 'argv' return execvp(file, argv); Message-Id: <20210321145118.27783-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Extract vsockserver-socketbinder and vsockserverdAlyssa Ross2021-03-21
| | | | | | | | | | | | | | | | | | | | | vsockserver-socketbinder creates and listens on a socket, and vsockserverd accepts connections and sets up file descriptors. vsockserver previously did both of these things in one big program, but now it just sets up a command line to run vsockserver-socketbinder followed by vsockserverd. Having two seperate programs allows one program to be used in situations where the other is not suitable (e.g. using vsockserver-socketbinder to create a socket in situations where accept behaviour more complex than vsockserverd can provide is required). This design is taken from s6[1], which uses the same design for its s6-ipcserver, s6-ipcserver-socketbinder, and s6-ipcserverd programs. [1]: https://skarnet.org/software/s6/ Message-Id: <20210319031713.23600-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* exec: add execzp()Alyssa Ross2021-03-21
| | | | | Message-Id: <20210319025648.17925-6-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* log: add die functionAlyssa Ross2021-03-21
| | | | | | | | This is useful if we want to provide a more specific error message than strerror can provide. Message-Id: <20210319025648.17925-5-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* configure: create, to generate config.hAlyssa Ross2021-03-21
| | | | | | | | | This will allow programs to refer to BINDIR to find other ucspi-vsock programs, which allows programs to be implemented in terms of each other. Message-Id: <20210319025648.17925-4-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsockserver: switch to a non-blocking socketAlyssa Ross2021-03-21
| | | | | | | | | | | | | | | This is entirely an implementation detail, for the moment, but it becomes important when we split out vsockserver into two seperate programs, one for binding the socket, and one for accepting, like s6-networking. Non-blocking sockets are the default there, so it makes sense to share that behaviour. I added the comment about blockingness of the connection sockets just in case, but for now Linux is the only Unix with an AF_VSOCK implementation, so I don't expect it to become relevant any time soon. Message-Id: <20210319025648.17925-3-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsock: check socket family before reading sockaddrAlyssa Ross2021-03-21
| | | | | | | | | Extracting a helper function for this has the nice side effect of making the `cid' and `port' parameters to vsock_accept nullable, which is nice for consistency with vsock_get_cid_and_port. Message-Id: <20210319025648.17925-2-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsock: get cid and port instead of just cidAlyssa Ross2021-03-21
| | | | | | | | We can get both in the same system call, so we might as well, rather than having two different functions that make the system call twice. Message-Id: <20210319025648.17925-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Makefile: use $@ where possibleAlyssa Ross2021-03-18
| | | | | | | | | This is, unfortunately, the only "internal macro" POSIX requires to be available for target rules, and I do want to keep the Makefile POSIX-compatible. But this is at least a slight improvement. Message-Id: <20210318190451.15182-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Makefile: stop repeating program listAlyssa Ross2021-03-18
| | | | | | | | | Using a wildcard with rm -f means that if there are no object files to remove, rm will try to remove a non-existent file named '*.o', and then ignore the failure it gets by trying to do that. Which is fine. Message-Id: <20210318003604.13621-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* num: rename from utilAlyssa Ross2021-03-18
| | | | | | | | I don't want to have anything called "util", because that will inevitably become a dumping ground for all sorts of unrelated stuff. Message-Id: <20210318003545.13414-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Factor out set_verbosity()Alyssa Ross2021-03-18
| | | | | | | This will save some cognitive load when reading main functions. Message-Id: <20210318003530.13208-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsockserver: consistently use {STDIN,STDOUT}_FILENOAlyssa Ross2021-03-18
| | | | | Message-Id: <20210318003505.12895-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsockserver: consistently use diee instead of errAlyssa Ross2021-03-18
| | | | | | | diee is our function that respects the verbosity setting. Message-Id: <20210318001256.4078-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Use C11 noreturn instead of GNU attribute extensionAlyssa Ross2021-03-18
| | | | | Message-Id: <20210318001240.3897-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Makefile: remove vsock{server,client}.o in cleanAlyssa Ross2021-03-18
| | | | | Message-Id: <20210318001129.2891-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Strip trailing whitespaceAlyssa Ross2021-03-18
|
* Makefile: link object files instead of source filesAlyssa Ross2021-03-10
| | | | | | | | | | | | Previously this would (I assume) needlessly recompile vsockserver.c and vsockclient.c when Make has already automatically compiled those. GNU Make has an automatic variable, $+, that we could use here, but currently the Makefile is portable, and it would be a shame to require GNU Make just for that. Message-Id: <20210310204555.20725-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsock_accept: return fd instead of 0 on successAlyssa Ross2021-03-10
| | | | | | | | | This would result in the spawned process being hooked up to stdin, instead of the vsock. Then stdin would be closed, so subsequent processes would be connected to nothing. Oops. Message-Id: <20210310204516.20041-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsock_open: return fd instead of 0 on successAlyssa Ross2021-03-10
| | | | | | | | | Previously, this returned the return value of connect(2), which is 0 on success. So stdin would be assigned to descriptors 6 and 7 in vsockclient instead of the socket. Oops. Message-Id: <20210309171816.8589-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* vsockserver: use fclose instead of closeAlyssa Ross2021-03-10
| | | | | | | | stdio can buffer output, so if we close stdout without going through stdio, there might be buffered output that is never written. Message-Id: <20210309154048.14474-1-hi@alyssa.is> Reviewed-by: Cole Helbling <cole.e.helbling@outlook.com>
* Initial commitAlyssa Ross2020-09-10
generated by cgit-pink 1.4.1 (git 2.36.1) at 2024-05-04 22:37:58 +0000