| Message ID | 20201102134136.2565-1-shipujin.t@gmail.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | net: ethernet: mscc: fix missing brace warning for old compilers | expand |
On Mon, Nov 02, 2020 at 09:41:36PM +0800, Pujin Shi wrote: > For older versions of gcc, the array = {0}; will cause warnings: > > drivers/net/ethernet/mscc/ocelot_vcap.c: In function 'is1_entry_set': > drivers/net/ethernet/mscc/ocelot_vcap.c:755:11: warning: missing braces around initializer [-Wmissing-braces] > struct ocelot_vcap_u16 etype = {0}; > ^ > drivers/net/ethernet/mscc/ocelot_vcap.c:755:11: warning: (near initialization for 'etype.value') [-Wmissing-braces] > > 1 warnings generated > > Fixes: 75944fda1dfe ("net: mscc: ocelot: offload ingress skbedit and vlan actions to VCAP IS1") > Signed-off-by: Pujin Shi <shipujin.t@gmail.com> > --- > drivers/net/ethernet/mscc/ocelot_vcap.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/net/ethernet/mscc/ocelot_vcap.c b/drivers/net/ethernet/mscc/ocelot_vcap.c > index d8c778ee6f1b..b96eab4583e7 100644 > --- a/drivers/net/ethernet/mscc/ocelot_vcap.c > +++ b/drivers/net/ethernet/mscc/ocelot_vcap.c > @@ -752,7 +752,7 @@ static void is1_entry_set(struct ocelot *ocelot, int ix, > dport); > } else { > /* IPv4 "other" frame */ > - struct ocelot_vcap_u16 etype = {0}; > + struct ocelot_vcap_u16 etype = {}; > > /* Overloaded field */ > etype.value[0] = proto.value[0]; > -- > 2.18.1 > Sorry, I don't understand what the problem is, or why your patch fixes it. What version of gcc are you testing with?
On Mon, Nov 02, 2020 at 01:56:55PM +0000, Vladimir Oltean wrote: > On Mon, Nov 02, 2020 at 09:41:36PM +0800, Pujin Shi wrote: > > For older versions of gcc, the array = {0}; will cause warnings: > > > > drivers/net/ethernet/mscc/ocelot_vcap.c: In function 'is1_entry_set': > > drivers/net/ethernet/mscc/ocelot_vcap.c:755:11: warning: missing braces around initializer [-Wmissing-braces] > > struct ocelot_vcap_u16 etype = {0}; > > ^ > > drivers/net/ethernet/mscc/ocelot_vcap.c:755:11: warning: (near initialization for 'etype.value') [-Wmissing-braces] > > > > 1 warnings generated > > > > Fixes: 75944fda1dfe ("net: mscc: ocelot: offload ingress skbedit and vlan actions to VCAP IS1") > > Signed-off-by: Pujin Shi <shipujin.t@gmail.com> > > --- > > drivers/net/ethernet/mscc/ocelot_vcap.c | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/drivers/net/ethernet/mscc/ocelot_vcap.c b/drivers/net/ethernet/mscc/ocelot_vcap.c > > index d8c778ee6f1b..b96eab4583e7 100644 > > --- a/drivers/net/ethernet/mscc/ocelot_vcap.c > > +++ b/drivers/net/ethernet/mscc/ocelot_vcap.c > > @@ -752,7 +752,7 @@ static void is1_entry_set(struct ocelot *ocelot, int ix, > > dport); > > } else { > > /* IPv4 "other" frame */ > > - struct ocelot_vcap_u16 etype = {0}; > > + struct ocelot_vcap_u16 etype = {}; > > > > /* Overloaded field */ > > etype.value[0] = proto.value[0]; > > Sorry, I don't understand what the problem is, or why your patch fixes > it. What version of gcc are you testing with? Nothing wrong and { 0 } is the right initializer, the reports must be from some ancient gcc but we weren't told which one either. https://lore.kernel.org/linux-btrfs/fbddb15a-6e46-3f21-23ba-b18f66e3448a@suse.com/
On Mon, 2 Nov 2020 13:56:55 +0000 Vladimir Oltean wrote: > On Mon, Nov 02, 2020 at 09:41:36PM +0800, Pujin Shi wrote: > > For older versions of gcc, the array = {0}; will cause warnings: > > > > drivers/net/ethernet/mscc/ocelot_vcap.c: In function 'is1_entry_set': > > drivers/net/ethernet/mscc/ocelot_vcap.c:755:11: warning: missing braces around initializer [-Wmissing-braces] > > struct ocelot_vcap_u16 etype = {0}; > > ^ > > drivers/net/ethernet/mscc/ocelot_vcap.c:755:11: warning: (near initialization for 'etype.value') [-Wmissing-braces] > > > > 1 warnings generated > > > > Fixes: 75944fda1dfe ("net: mscc: ocelot: offload ingress skbedit and vlan actions to VCAP IS1") > > Signed-off-by: Pujin Shi <shipujin.t@gmail.com> > > --- > > drivers/net/ethernet/mscc/ocelot_vcap.c | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/drivers/net/ethernet/mscc/ocelot_vcap.c b/drivers/net/ethernet/mscc/ocelot_vcap.c > > index d8c778ee6f1b..b96eab4583e7 100644 > > --- a/drivers/net/ethernet/mscc/ocelot_vcap.c > > +++ b/drivers/net/ethernet/mscc/ocelot_vcap.c > > @@ -752,7 +752,7 @@ static void is1_entry_set(struct ocelot *ocelot, int ix, > > dport); > > } else { > > /* IPv4 "other" frame */ > > - struct ocelot_vcap_u16 etype = {0}; > > + struct ocelot_vcap_u16 etype = {}; > > > > Sorry, I don't understand what the problem is, or why your patch fixes > it. What version of gcc are you testing with? Old GCC does not like the 0, if the members of struct are not scalars. struct ocelot_vcap_u16 { u8 value[2]; u8 mask[2]; }; In this case the first member is an array. It wants us to add another curly brace: struct ocelot_vcap_u16 etype = {{0}}; ... or we can just skip the 0. That's just FWIW. I don't remember which versions of GCC behave like that, I just know we get a constant stream of this sort of fixes. I think clang may generate a similar warning. Pujin, please specify the version of GCC you're using and repost.
On Mon, Nov 02, 2020 at 02:08:08PM -0800, Jakub Kicinski wrote: > Old GCC does not like the 0, if the members of struct are not scalars. > > struct ocelot_vcap_u16 { > u8 value[2]; > u8 mask[2]; > }; > > In this case the first member is an array. > > It wants us to add another curly brace: > > struct ocelot_vcap_u16 etype = {{0}}; > > ... or we can just skip the 0. I am reading that the empty set initializer is a GNU extension. I would not like to see yet another patch from clang folks coming as a follow-up to this one. > That's just FWIW. I don't remember which versions of GCC behave like > that, I just know we get a constant stream of this sort of fixes. > I think clang may generate a similar warning. > > Pujin, please specify the version of GCC you're using and repost. If the initializer really has to be changed, I would prefer to see a memset.
diff --git a/drivers/net/ethernet/mscc/ocelot_vcap.c b/drivers/net/ethernet/mscc/ocelot_vcap.c index d8c778ee6f1b..b96eab4583e7 100644 --- a/drivers/net/ethernet/mscc/ocelot_vcap.c +++ b/drivers/net/ethernet/mscc/ocelot_vcap.c @@ -752,7 +752,7 @@ static void is1_entry_set(struct ocelot *ocelot, int ix, dport); } else { /* IPv4 "other" frame */ - struct ocelot_vcap_u16 etype = {0}; + struct ocelot_vcap_u16 etype = {}; /* Overloaded field */ etype.value[0] = proto.value[0];
For older versions of gcc, the array = {0}; will cause warnings: drivers/net/ethernet/mscc/ocelot_vcap.c: In function 'is1_entry_set': drivers/net/ethernet/mscc/ocelot_vcap.c:755:11: warning: missing braces around initializer [-Wmissing-braces] struct ocelot_vcap_u16 etype = {0}; ^ drivers/net/ethernet/mscc/ocelot_vcap.c:755:11: warning: (near initialization for 'etype.value') [-Wmissing-braces] 1 warnings generated Fixes: 75944fda1dfe ("net: mscc: ocelot: offload ingress skbedit and vlan actions to VCAP IS1") Signed-off-by: Pujin Shi <shipujin.t@gmail.com> --- drivers/net/ethernet/mscc/ocelot_vcap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)