Re: [PATCH 4/4] tools/misc: Test for fault injection

[Jan Beulich <jbeulich@xxxxxxxx>]

On 23.12.2020 17:34, Andrew Cooper wrote:
> --- /dev/null
> +++ b/tools/misc/xen-fault-ttl.c
> @@ -0,0 +1,56 @@
> +#include <stdio.h>
> +#include <err.h>
> +#include <string.h>
> +#include <errno.h>
> +
> +#include <xenctrl.h>
> +
> +int main(int argc, char **argv)
> +{
> +    int rc;
> +    xc_interface *xch = xc_interface_open(NULL, NULL, 0);
> +
> +    if ( !xch )
> +        err(1, "xc_interface_open");
> +
> +    struct xen_domctl_createdomain create = {
> +        .fault_ttl = 1,
> +        .flags = (XEN_DOMCTL_CDF_hvm |
> +                  XEN_DOMCTL_CDF_hap),
> +        .max_vcpus = 1,
> +        .max_evtchn_port = -1,
> +        .max_grant_frames = 64,
> +        .max_maptrack_frames = 1024,
> +        .arch = {
> +            .emulation_flags = XEN_X86_EMU_LAPIC,
> +        },
> +    };
> +    uint32_t domid = 0;
> +
> +    for ( rc = 1; rc && errno == ENOMEM; create.fault_ttl++ )
> +        rc = xc_domain_create(xch, &domid, &create);

The very first time you get here, how does errno end up being
set to ENOMEM? To me

    do {
        create.fault_ttl++;
        rc = xc_domain_create(xch, &domid, &create);
    } while ( rc && errno == ENOMEM );

would seem more natural (and also avoid the somewhat odd
"rc = 1").

Jan