[PATCH] libqtest: refuse QTEST_QEMU_BINARY=qemu-kvm

Stefan Hajnoczi posted 1 patch 6 days, 20 hours ago
Test checkpatch passed
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20210407104511.343061-1-stefanha@redhat.com
tests/qtest/libqtest.c | 8 ++++++++
1 file changed, 8 insertions(+)

[PATCH] libqtest: refuse QTEST_QEMU_BINARY=qemu-kvm

Posted by Stefan Hajnoczi 6 days, 20 hours ago
Some downstreams rename the QEMU binary to "qemu-kvm". This breaks
qtest_get_arch(), which attempts to parse the target architecture from
the QTEST_QEMU_BINARY environment variable.

Print an error instead of returning the architecture "kvm". Things fail
in weird ways when the architecture string is bogus.

Arguably qtests should always be run in a build directory instead of
against an installed QEMU. In any case, printing a clear error when this
happens is helpful.

Reported-by: Qin Wang <qinwang@rehdat.com>
Cc: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/qtest/libqtest.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/tests/qtest/libqtest.c b/tests/qtest/libqtest.c
index 71e359efcd..2fc249c0b5 100644
--- a/tests/qtest/libqtest.c
+++ b/tests/qtest/libqtest.c
@@ -910,6 +910,14 @@ const char *qtest_get_arch(void)
         abort();
     }
 
+    if (strcmp(end + 1, "kvm") == 0) {
+        fprintf(stderr, "QTEST_QEMU_BINARY must end with *-<arch>. If you are "
+                        "using qemu-kvm, please create a symlink like ln -s "
+                        "path/to/qemu-kvm qemu-system-x86_64 and use that "
+                        "instead.\n");
+        abort();
+    }
+
     return end + 1;
 }
 
-- 
2.30.2

Re: [PATCH] libqtest: refuse QTEST_QEMU_BINARY=qemu-kvm

Posted by Peter Maydell 5 days, 20 hours ago
On Wed, 7 Apr 2021 at 11:46, Stefan Hajnoczi <stefanha@redhat.com> wrote:
>
> Some downstreams rename the QEMU binary to "qemu-kvm". This breaks
> qtest_get_arch(), which attempts to parse the target architecture from
> the QTEST_QEMU_BINARY environment variable.
>
> Print an error instead of returning the architecture "kvm". Things fail
> in weird ways when the architecture string is bogus.
>
> Arguably qtests should always be run in a build directory instead of
> against an installed QEMU. In any case, printing a clear error when this
> happens is helpful.
>
> Reported-by: Qin Wang <qinwang@rehdat.com>
> Cc: Emanuele Giuseppe Esposito <eesposit@redhat.com>
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> ---
>  tests/qtest/libqtest.c | 8 ++++++++
>  1 file changed, 8 insertions(+)

This way of figuring out the architecture has always been a bit cheesy.
It makes it awkward for instance if you want to put a wrapper script
around the qemu binary (to enable debug options, for instance) -- you
have to be careful to name the script with the right suffix.

thanks
-- PMM

Re: [PATCH] libqtest: refuse QTEST_QEMU_BINARY=qemu-kvm

Posted by Laurent Vivier 5 days, 20 hours ago
On 07/04/2021 12:45, Stefan Hajnoczi wrote:
> Some downstreams rename the QEMU binary to "qemu-kvm". This breaks
> qtest_get_arch(), which attempts to parse the target architecture from
> the QTEST_QEMU_BINARY environment variable.
> 
> Print an error instead of returning the architecture "kvm". Things fail
> in weird ways when the architecture string is bogus.
> 
> Arguably qtests should always be run in a build directory instead of
> against an installed QEMU. In any case, printing a clear error when this
> happens is helpful.
> 
> Reported-by: Qin Wang <qinwang@rehdat.com>
> Cc: Emanuele Giuseppe Esposito <eesposit@redhat.com>
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> ---
>  tests/qtest/libqtest.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/tests/qtest/libqtest.c b/tests/qtest/libqtest.c
> index 71e359efcd..2fc249c0b5 100644
> --- a/tests/qtest/libqtest.c
> +++ b/tests/qtest/libqtest.c
> @@ -910,6 +910,14 @@ const char *qtest_get_arch(void)
>          abort();
>      }
>  
> +    if (strcmp(end + 1, "kvm") == 0) {
> +        fprintf(stderr, "QTEST_QEMU_BINARY must end with *-<arch>. If you are "
> +                        "using qemu-kvm, please create a symlink like ln -s "
> +                        "path/to/qemu-kvm qemu-system-x86_64 and use that "
> +                        "instead.\n");
> +        abort();
> +    }
> +
>      return end + 1;
>  }
>  

I think it would be more generic to check we have the "-system-" string in the name.

Thanks,
Laurent


Re: [PATCH] libqtest: refuse QTEST_QEMU_BINARY=qemu-kvm

Posted by Stefan Hajnoczi 5 days, 15 hours ago
On Thu, Apr 08, 2021 at 12:37:36PM +0200, Laurent Vivier wrote:
> On 07/04/2021 12:45, Stefan Hajnoczi wrote:
> > Some downstreams rename the QEMU binary to "qemu-kvm". This breaks
> > qtest_get_arch(), which attempts to parse the target architecture from
> > the QTEST_QEMU_BINARY environment variable.
> > 
> > Print an error instead of returning the architecture "kvm". Things fail
> > in weird ways when the architecture string is bogus.
> > 
> > Arguably qtests should always be run in a build directory instead of
> > against an installed QEMU. In any case, printing a clear error when this
> > happens is helpful.
> > 
> > Reported-by: Qin Wang <qinwang@rehdat.com>
> > Cc: Emanuele Giuseppe Esposito <eesposit@redhat.com>
> > Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> > ---
> >  tests/qtest/libqtest.c | 8 ++++++++
> >  1 file changed, 8 insertions(+)
> > 
> > diff --git a/tests/qtest/libqtest.c b/tests/qtest/libqtest.c
> > index 71e359efcd..2fc249c0b5 100644
> > --- a/tests/qtest/libqtest.c
> > +++ b/tests/qtest/libqtest.c
> > @@ -910,6 +910,14 @@ const char *qtest_get_arch(void)
> >          abort();
> >      }
> >  
> > +    if (strcmp(end + 1, "kvm") == 0) {
> > +        fprintf(stderr, "QTEST_QEMU_BINARY must end with *-<arch>. If you are "
> > +                        "using qemu-kvm, please create a symlink like ln -s "
> > +                        "path/to/qemu-kvm qemu-system-x86_64 and use that "
> > +                        "instead.\n");
> > +        abort();
> > +    }
> > +
> >      return end + 1;
> >  }
> >  
> 
> I think it would be more generic to check we have the "-system-" string in the name.

Nice. I will send a v2.

Stefan