| Message ID | 20210813073615.32837-9-frankja@linux.ibm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | s390x: Cleanup and maintenance | expand |
On Fri, 13 Aug 2021 07:36:15 +0000 Janosch Frank <frankja@linux.ibm.com> wrote: > Let's not get bitten by an extension of the query struct and handle > the rc 0x100 error properly which does indicate that the UV has more > data for us. > > Signed-off-by: Janosch Frank <frankja@linux.ibm.com> Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com> > --- > lib/s390x/uv.c | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) > > diff --git a/lib/s390x/uv.c b/lib/s390x/uv.c > index fd9de944..c5c69c47 100644 > --- a/lib/s390x/uv.c > +++ b/lib/s390x/uv.c > @@ -49,6 +49,8 @@ int uv_setup(void) > if (!test_facility(158)) > return 0; > > - assert(!uv_call(0, (u64)&uvcb_qui)); > + uv_call(0, (u64)&uvcb_qui); > + > + assert(uvcb_qui.header.rc == 1 || uvcb_qui.header.rc == > 0x100); return 1; > }
On 13/08/2021 09.36, Janosch Frank wrote: > Let's not get bitten by an extension of the query struct and handle > the rc 0x100 error properly which does indicate that the UV has more > data for us. > > Signed-off-by: Janosch Frank <frankja@linux.ibm.com> > --- > lib/s390x/uv.c | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) > > diff --git a/lib/s390x/uv.c b/lib/s390x/uv.c > index fd9de944..c5c69c47 100644 > --- a/lib/s390x/uv.c > +++ b/lib/s390x/uv.c > @@ -49,6 +49,8 @@ int uv_setup(void) > if (!test_facility(158)) > return 0; > > - assert(!uv_call(0, (u64)&uvcb_qui)); > + uv_call(0, (u64)&uvcb_qui); > + > + assert(uvcb_qui.header.rc == 1 || uvcb_qui.header.rc == 0x100); Don't you want to continue to check the return code of the uv_call() function? Thomas
On 8/18/21 11:30 AM, Thomas Huth wrote: > On 13/08/2021 09.36, Janosch Frank wrote: >> Let's not get bitten by an extension of the query struct and handle >> the rc 0x100 error properly which does indicate that the UV has more >> data for us. >> >> Signed-off-by: Janosch Frank <frankja@linux.ibm.com> >> --- >> lib/s390x/uv.c | 4 +++- >> 1 file changed, 3 insertions(+), 1 deletion(-) >> >> diff --git a/lib/s390x/uv.c b/lib/s390x/uv.c >> index fd9de944..c5c69c47 100644 >> --- a/lib/s390x/uv.c >> +++ b/lib/s390x/uv.c >> @@ -49,6 +49,8 @@ int uv_setup(void) >> if (!test_facility(158)) >> return 0; >> >> - assert(!uv_call(0, (u64)&uvcb_qui)); >> + uv_call(0, (u64)&uvcb_qui); >> + >> + assert(uvcb_qui.header.rc == 1 || uvcb_qui.header.rc == 0x100); > > Don't you want to continue to check the return code of the uv_call() function? > > Thomas > The rc==0x100 case is a cc==1 and the rc==1 is a cc==0 so I had to delete the check. Those smaller patches are already upstream btw.
diff --git a/lib/s390x/uv.c b/lib/s390x/uv.c index fd9de944..c5c69c47 100644 --- a/lib/s390x/uv.c +++ b/lib/s390x/uv.c @@ -49,6 +49,8 @@ int uv_setup(void) if (!test_facility(158)) return 0; - assert(!uv_call(0, (u64)&uvcb_qui)); + uv_call(0, (u64)&uvcb_qui); + + assert(uvcb_qui.header.rc == 1 || uvcb_qui.header.rc == 0x100); return 1; }
Let's not get bitten by an extension of the query struct and handle the rc 0x100 error properly which does indicate that the UV has more data for us. Signed-off-by: Janosch Frank <frankja@linux.ibm.com> --- lib/s390x/uv.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)