Jim Meyering wrote:
Dave Allan <dallan@redhat.com> wrote:
Richard W.M. Jones wrote:
On Fri, Jan 23, 2009 at 02:51:02PM -0500, Dave Allan wrote:
The examples directory doesn't have a trivial example of how to connect to a hypervisor, make a few calls, and disconnect, so I put one together. I would appreciate any suggestions on anything that I've done wrong as well as suggestions for other fundamental API calls that should be illustrated. Yes, I checked this example code and it is fine. My only comment would be on:
+ /* virConnectOpenAuth called here with all default parameters */ + conn = virConnectOpenAuth(NULL, virConnectAuthPtrDefault, 0); It might be better to let people connect to a named URI.
Another possibility is to default to the test URI (test:///default) since that (almost) always exists. Hi Rich,
Thanks for taking a look at it. I added a little code to let the user specify a URI on the command line. Do you think it is worth committing?
Hi Dave,
I like your example. Thanks for preparing it.
Here are some suggestions:
Thanks for the style suggestions--that's one of the reasons I was sending the code around.
diff --git a/examples/hellolibvirt/hellolibvirt.c b/examples/hellolibvirt/hellolibvirt.c new file mode 100644 index 0000000..22d3309 --- /dev/null +++ b/examples/hellolibvirt/hellolibvirt.c @@ -0,0 +1,151 @@ +/* This file contains trivial example code to connect to the running + * hypervisor and gather a few bits of information. */ + +#include <stdio.h> +#include <stdlib.h> +#include <libvirt/libvirt.h> + +static int +showHypervisorInfo(virConnectPtr conn) +{ + int ret = 0; + unsigned long hvVer, major, minor, release; + const char *hvType; + + /* virConnectGetType returns a pointer to a static string, so no + * allocation or freeing is necessary; it is possible for the call + * to fail if, for example, there is no connection to a + * hypervisor, so check what it returns. */ + hvType = virConnectGetType(conn); + if (NULL == hvType) { + ret = 1; + printf("Failed to get hypervisor type\n"); + goto out; + } + + if (0 != virConnectGetVersion(conn, &hvVer)) { + ret = 1; + printf("Failed to get hypervisor version\n"); + goto out; + } + + major = hvVer / 1000000; + hvVer %= 1000000; + minor = hvVer / 1000; + release = hvVer % 1000; + + printf("Hypervisor: \"%s\" version: %lu.%lu.%lu\n", + hvType, + major, + minor, + release); +
How about initializing ret = 1 above and setting ret = 0 here to indicate success? It's a close call, since that results in removal of only two "ret = 1" assignments.
In this case, I think that the error cases are very unlikely, so I made the initialization 0, but I agree, it could go either way. I left it as is for now.
+out: + return ret; +} + + +static int +showDomains(virConnectPtr conn) +{ + int ret = 0, i, numNames, numInactiveDomains, numActiveDomains; + char **nameList = NULL; + + numActiveDomains = virConnectNumOfDomains(conn); + numInactiveDomains = virConnectNumOfDefinedDomains(conn);
It'd be good to handle numInactiveDomains < 0 differently. Currently it'll probably provoke a failed malloc, below.
Doh--thanks. I missed that those calls could fail.
+ printf("There are %d active and %d inactive domains\n", + numActiveDomains, numInactiveDomains); + + nameList = malloc(sizeof(char *) * (unsigned int)numInactiveDomains);
Using the target variable name rather than the type is a little more maintainable, in general, so set a good example: And please drop the cast. We hate casts, and besides, it's not needed. nameList = malloc(sizeof(*nameList) * numInactiveDomains);
Thanks on sizeof(char *) vs. sizeof(*nameList)--fixed. The cast was there because virConnectNumOfDefinedDomains returns a signed value to allow for returning -1 on error, but malloc expects an unsigned argument. gcc 4.3 -Wconversion complains about this situation [different behavior from gcc < 4.3] I've turned off that warning and removed the cast.
+ if (NULL == nameList) { + ret = 1; + printf("Could not allocate memory for list of inactive domains\n"); + goto out; + } + + numNames = virConnectListDefinedDomains(conn, + nameList, + numInactiveDomains); + + if (-1 == numNames) { + ret = 1; + printf("Could not get list of defined domains from hypervisor\n"); + goto out; + } + + if (numNames > 0) { + printf("Inactive domains:\n"); + } + + for (i = 0 ; i < numNames ; i++) { + printf(" %s\n", *(nameList + i)); + /* The API documentation doesn't say so, but the names + * returned by virConnectListDefinedDomains are strdup'd and + * must be freed here. */ + free(*(nameList + i)); + }
Here's another case where you can save a line by initializing ret=1 up front and setting ret=0 here.
+out: + if (NULL != nameList) { + free(nameList);
The test for non-NULL-before-free is unnecessary, since free is guaranteed to handle NULL properly. So just call free:
free(nameList);
In fact, if you run "make syntax-check" before making the suggested change, it should detect and complain about this code.
Removed. (make syntax-check does not complain, btw)
+ return ret; +} + + +int +main(int argc, char *argv[]) +{ + int ret = 0; + virConnectPtr conn = NULL;
The above initialization is unnecessary.
Fixed.
+ char *uri = NULL;
This one can be adjusted, too:
+ printf("Attempting to connect to hypervisor\n"); + + if (argc > 0) { + uri = argv[1]; + }
I'd write it as follows,
char *uri = (argc > 0 ? argv[1] : NULL);
so that it's clear the variable is defined unconditionally.
I tend not to use the ternary operator much, because I've seen it abused to write really obfuscated code, but you're right, this is a place where it makes things cleaner. Done.
In libvirt, it's ok to use C99 declaration-after-stmt.
Good to know.
+ /* virConnectOpenAuth is called here with all default parameters, + * except, possibly, the URI of the hypervisor. */ + conn = virConnectOpenAuth(uri, virConnectAuthPtrDefault, 0); + + if (NULL == conn) { + ret = 1; + printf("No connection to hypervisor\n"); + goto out; + } + + uri = virConnectGetURI(conn); + if (NULL == uri) { + ret = 1; + printf("Failed to get URI for hypervisor connection\n"); + goto disconnect; + } + + printf("Connected to hypervisor at \"%s\"\n", uri); + free(uri); + + if (0 != showHypervisorInfo(conn)) { + ret = 1; + goto disconnect; + } + + if (0 != showDomains(conn)) { + ret = 1; + goto disconnect; + } + +disconnect: + if (0 != virConnectClose(conn)) { + printf("Failed to disconnect from hypervisor\n"); + } else { + printf("Disconnected from hypervisor\n"); + }
You can save 3 statements by hoisting the initialization of ret=1 and setting ret=0 here.
+out: + return ret; +}
I noticed that you're using the git mirror. Good! ;-) When posting patches, please use "git format-patch".
Will do.
That would have made it easier for me to apply and test your patches. As it is, I didn't do either because "git am FILE" didn't work:
$ git am dallan.patch Applying: trivial libvirt example code warning: examples/hellolibvirt/hellolibvirt.c has type 100755, expected 100644 error: patch failed: examples/hellolibvirt/hellolibvirt.c:97 error: examples/hellolibvirt/hellolibvirt.c: patch does not apply Patch failed at 0001 trivial libvirt example code When you have resolved this problem run "git am --resolved". If you would prefer to skip this patch, instead run "git am --skip". To restore the original branch and stop patching run "git am --abort".
Note the warning about permissions on hellolibvirt.c. You can correct that by running "chmod a-x hellolibvirt.c".
The permissions problem is strange--it's 644 in my development tree, and the patch I sent has: diff --git a/examples/hellolibvirt/hellolibvirt.c b/examples/hellolibvirt/hellolibvirt.c new file mode 100644 What would cause git-am to think it was 755?
Here are some contribution guidelines that generally make it easier for maintainers/committers to deal with contributed patches, (though some parts are specific to git-managed projects):
http://git.sv.gnu.org/gitweb/?p=coreutils.git;a=blob;f=HACKING;hb=HEAD
Good stuff. When I have a patch like this that people have commented on and I've modified it slightly in response, what's the best way to re-submit it? When Rich responded, I submitted both the entire patch with the changes as well as the changes separately. I'll resend a patch when I've gotten git to squash the history properly to produce something usable from git-format-patch. Dave