| Message ID | 20220711212932.1501592-1-justinstitt@google.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 68204a696505fe97150d498f32e38c2a926c540f |
| Delegated to: | Kalle Valo |
| Headers | show |
| Series | mediatek: mt7601u: fix clang -Wformat warning | expand |
On Mon, 11 Jul 2022 14:29:32 -0700 Justin Stitt wrote: > When building with Clang we encounter this warning: > | drivers/net/wireless/mediatek/mt7601u/debugfs.c:92:6: error: format > | specifies type 'unsigned char' but the argument has type 'int' > | [-Werror,-Wformat] dev->ee->reg.start + dev->ee->reg.num - 1); > > The format specifier used is `%hhu` which describes a u8. Both > `dev->ee->reg.start` and `.num` are u8 as well. However, the expression > as a whole is promoted to an int as you cannot get smaller-than-int from > addition. Therefore, to fix the warning, use the promoted-to-type's > format specifier -- in this case `%d`. Acked-by: Jakub Kicinski <kubakici@wp.pl>
On Mon, 2022-07-11 at 14:29 -0700, Justin Stitt wrote: > When building with Clang we encounter this warning: > > drivers/net/wireless/mediatek/mt7601u/debugfs.c:92:6: error: format > > specifies type 'unsigned char' but the argument has type 'int' > > [-Werror,-Wformat] dev->ee->reg.start + dev->ee->reg.num - 1); > > The format specifier used is `%hhu` which describes a u8. Both > `dev->ee->reg.start` and `.num` are u8 as well. However, the expression > as a whole is promoted to an int as you cannot get smaller-than-int from > addition. Therefore, to fix the warning, use the promoted-to-type's > format specifier -- in this case `%d`. I think whenever a sizeof(unsigned type) that is less than sizeof(int) is emitted with vsprintf, the preferred format specifier should be %u not %d. > diff --git a/drivers/net/wireless/mediatek/mt7601u/debugfs.c b/drivers/net/wireless/mediatek/mt7601u/debugfs.c [] > @@ -88,7 +88,7 @@ mt7601u_eeprom_param_show(struct seq_file *file, void *data) > dev->ee->rssi_offset[0], dev->ee->rssi_offset[1]); > seq_printf(file, "Reference temp: %hhx\n", dev->ee->ref_temp); > seq_printf(file, "LNA gain: %hhx\n", dev->ee->lna_gain); > - seq_printf(file, "Reg channels: %hhu-%hhu\n", dev->ee->reg.start, > + seq_printf(file, "Reg channels: %hhu-%d\n", dev->ee->reg.start, > dev->ee->reg.start + dev->ee->reg.num - 1); And this is not a promotion of an argument to int via varargs. The arithmetic did the promotion. I suggest s/%hh/%/ for all the uses here, not just this one. checkpatch could do this somewhat automatically. Of course any changes it suggests need human review. $ ./scripts/checkpatch.pl -f drivers/net/wireless/mediatek/mt7601u/debugfs.c --show-types --types=unnecessary_modifier --fix-inplace $ git diff --stat -p drivers/net/wireless/mediatek/mt7601u/debugfs.c --- drivers/net/wireless/mediatek/mt7601u/debugfs.c | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-) diff --git a/drivers/net/wireless/mediatek/mt7601u/debugfs.c b/drivers/net/wireless/mediatek/mt7601u/debugfs.c index 20669eacb66ea..b7a6376e3352e 100644 --- a/drivers/net/wireless/mediatek/mt7601u/debugfs.c +++ b/drivers/net/wireless/mediatek/mt7601u/debugfs.c @@ -83,28 +83,28 @@ mt7601u_eeprom_param_show(struct seq_file *file, void *data) struct tssi_data *td = &dev->ee->tssi_data; int i; - seq_printf(file, "RF freq offset: %hhx\n", dev->ee->rf_freq_off); - seq_printf(file, "RSSI offset: %hhx %hhx\n", + seq_printf(file, "RF freq offset: %x\n", dev->ee->rf_freq_off); + seq_printf(file, "RSSI offset: %x %x\n", dev->ee->rssi_offset[0], dev->ee->rssi_offset[1]); - seq_printf(file, "Reference temp: %hhx\n", dev->ee->ref_temp); - seq_printf(file, "LNA gain: %hhx\n", dev->ee->lna_gain); - seq_printf(file, "Reg channels: %hhu-%hhu\n", dev->ee->reg.start, + seq_printf(file, "Reference temp: %x\n", dev->ee->ref_temp); + seq_printf(file, "LNA gain: %x\n", dev->ee->lna_gain); + seq_printf(file, "Reg channels: %u-%u\n", dev->ee->reg.start, dev->ee->reg.start + dev->ee->reg.num - 1); seq_puts(file, "Per rate power:\n"); for (i = 0; i < 2; i++) - seq_printf(file, "\t raw:%02hhx bw20:%02hhx bw40:%02hhx\n", + seq_printf(file, "\t raw:%02x bw20:%02x bw40:%02x\n", rp->cck[i].raw, rp->cck[i].bw20, rp->cck[i].bw40); for (i = 0; i < 4; i++) - seq_printf(file, "\t raw:%02hhx bw20:%02hhx bw40:%02hhx\n", + seq_printf(file, "\t raw:%02x bw20:%02x bw40:%02x\n", rp->ofdm[i].raw, rp->ofdm[i].bw20, rp->ofdm[i].bw40); for (i = 0; i < 4; i++) - seq_printf(file, "\t raw:%02hhx bw20:%02hhx bw40:%02hhx\n", + seq_printf(file, "\t raw:%02x bw20:%02x bw40:%02x\n", rp->ht[i].raw, rp->ht[i].bw20, rp->ht[i].bw40); seq_puts(file, "Per channel power:\n"); for (i = 0; i < 7; i++) - seq_printf(file, "\t tx_power ch%u:%02hhx ch%u:%02hhx\n", + seq_printf(file, "\t tx_power ch%u:%02x ch%u:%02x\n", i * 2 + 1, dev->ee->chan_pwr[i * 2], i * 2 + 2, dev->ee->chan_pwr[i * 2 + 1]); @@ -112,8 +112,8 @@ mt7601u_eeprom_param_show(struct seq_file *file, void *data) return 0; seq_puts(file, "TSSI:\n"); - seq_printf(file, "\t slope:%02hhx\n", td->slope); - seq_printf(file, "\t offset=%02hhx %02hhx %02hhx\n", + seq_printf(file, "\t slope:%02x\n", td->slope); + seq_printf(file, "\t offset=%02x %02x %02x\n", td->offset[0], td->offset[1], td->offset[2]); seq_printf(file, "\t delta_off:%08x\n", td->tx0_delta_offset);
On Thu, Jul 14, 2022 at 11:33 PM Joe Perches <joe@perches.com> wrote: > > On Mon, 2022-07-11 at 14:29 -0700, Justin Stitt wrote: > > When building with Clang we encounter this warning: > > > drivers/net/wireless/mediatek/mt7601u/debugfs.c:92:6: error: format > > > specifies type 'unsigned char' but the argument has type 'int' > > > [-Werror,-Wformat] dev->ee->reg.start + dev->ee->reg.num - 1); > > > > The format specifier used is `%hhu` which describes a u8. Both > > `dev->ee->reg.start` and `.num` are u8 as well. However, the expression > > as a whole is promoted to an int as you cannot get smaller-than-int from > > addition. Therefore, to fix the warning, use the promoted-to-type's > > format specifier -- in this case `%d`. > > I think whenever a sizeof(unsigned type) that is less than sizeof(int) is > emitted with vsprintf, the preferred format specifier should be %u not %d. I believe the standard recommends using the promoted-to-type's format specifier, in this case integer. I agree, though, that it is quite bizarre to represent unsigned types with signed types. I'd be interested in hearing more opinions on the matter. > > diff --git a/drivers/net/wireless/mediatek/mt7601u/debugfs.c b/drivers/net/wireless/mediatek/mt7601u/debugfs.c > [] > > @@ -88,7 +88,7 @@ mt7601u_eeprom_param_show(struct seq_file *file, void *data) > > dev->ee->rssi_offset[0], dev->ee->rssi_offset[1]); > > seq_printf(file, "Reference temp: %hhx\n", dev->ee->ref_temp); > > seq_printf(file, "LNA gain: %hhx\n", dev->ee->lna_gain); > > - seq_printf(file, "Reg channels: %hhu-%hhu\n", dev->ee->reg.start, > > + seq_printf(file, "Reg channels: %hhu-%d\n", dev->ee->reg.start, > > dev->ee->reg.start + dev->ee->reg.num - 1); > > And this is not a promotion of an argument to int via varargs. > The arithmetic did the promotion. Right, I noted in my patch message that the type promotion was due to addition. > > I suggest s/%hh/%/ for all the uses here, not just this one. I also contemplated this change but I think it might be a bit out of scope for https://github.com/ClangBuiltLinux/linux/issues/378 -- What do you think? It could be argued that every single instance of %hh[dux] should be replaced with %[dux]. > checkpatch could do this somewhat automatically. > Of course any changes it suggests need human review. > > $ ./scripts/checkpatch.pl -f drivers/net/wireless/mediatek/mt7601u/debugfs.c --show-types --types=unnecessary_modifier --fix-inplace > $ git diff --stat -p drivers/net/wireless/mediatek/mt7601u/debugfs.c > --- > drivers/net/wireless/mediatek/mt7601u/debugfs.c | 22 +++++++++++----------- > 1 file changed, 11 insertions(+), 11 deletions(-) > > diff --git a/drivers/net/wireless/mediatek/mt7601u/debugfs.c b/drivers/net/wireless/mediatek/mt7601u/debugfs.c > index 20669eacb66ea..b7a6376e3352e 100644 > --- a/drivers/net/wireless/mediatek/mt7601u/debugfs.c > +++ b/drivers/net/wireless/mediatek/mt7601u/debugfs.c > @@ -83,28 +83,28 @@ mt7601u_eeprom_param_show(struct seq_file *file, void *data) > struct tssi_data *td = &dev->ee->tssi_data; > int i; > > - seq_printf(file, "RF freq offset: %hhx\n", dev->ee->rf_freq_off); > - seq_printf(file, "RSSI offset: %hhx %hhx\n", > + seq_printf(file, "RF freq offset: %x\n", dev->ee->rf_freq_off); > + seq_printf(file, "RSSI offset: %x %x\n", > dev->ee->rssi_offset[0], dev->ee->rssi_offset[1]); > - seq_printf(file, "Reference temp: %hhx\n", dev->ee->ref_temp); > - seq_printf(file, "LNA gain: %hhx\n", dev->ee->lna_gain); > - seq_printf(file, "Reg channels: %hhu-%hhu\n", dev->ee->reg.start, > + seq_printf(file, "Reference temp: %x\n", dev->ee->ref_temp); > + seq_printf(file, "LNA gain: %x\n", dev->ee->lna_gain); > + seq_printf(file, "Reg channels: %u-%u\n", dev->ee->reg.start, > dev->ee->reg.start + dev->ee->reg.num - 1); > > seq_puts(file, "Per rate power:\n"); > for (i = 0; i < 2; i++) > - seq_printf(file, "\t raw:%02hhx bw20:%02hhx bw40:%02hhx\n", > + seq_printf(file, "\t raw:%02x bw20:%02x bw40:%02x\n", > rp->cck[i].raw, rp->cck[i].bw20, rp->cck[i].bw40); > for (i = 0; i < 4; i++) > - seq_printf(file, "\t raw:%02hhx bw20:%02hhx bw40:%02hhx\n", > + seq_printf(file, "\t raw:%02x bw20:%02x bw40:%02x\n", > rp->ofdm[i].raw, rp->ofdm[i].bw20, rp->ofdm[i].bw40); > for (i = 0; i < 4; i++) > - seq_printf(file, "\t raw:%02hhx bw20:%02hhx bw40:%02hhx\n", > + seq_printf(file, "\t raw:%02x bw20:%02x bw40:%02x\n", > rp->ht[i].raw, rp->ht[i].bw20, rp->ht[i].bw40); > > seq_puts(file, "Per channel power:\n"); > for (i = 0; i < 7; i++) > - seq_printf(file, "\t tx_power ch%u:%02hhx ch%u:%02hhx\n", > + seq_printf(file, "\t tx_power ch%u:%02x ch%u:%02x\n", > i * 2 + 1, dev->ee->chan_pwr[i * 2], > i * 2 + 2, dev->ee->chan_pwr[i * 2 + 1]); > > @@ -112,8 +112,8 @@ mt7601u_eeprom_param_show(struct seq_file *file, void *data) > return 0; > > seq_puts(file, "TSSI:\n"); > - seq_printf(file, "\t slope:%02hhx\n", td->slope); > - seq_printf(file, "\t offset=%02hhx %02hhx %02hhx\n", > + seq_printf(file, "\t slope:%02x\n", td->slope); > + seq_printf(file, "\t offset=%02x %02x %02x\n", > td->offset[0], td->offset[1], td->offset[2]); > seq_printf(file, "\t delta_off:%08x\n", td->tx0_delta_offset); > >
On Fri, 2022-07-15 at 17:08 -0700, Justin Stitt wrote: > On Thu, Jul 14, 2022 at 11:33 PM Joe Perches <joe@perches.com> wrote: > > > > On Mon, 2022-07-11 at 14:29 -0700, Justin Stitt wrote: > > > When building with Clang we encounter this warning: > > > > drivers/net/wireless/mediatek/mt7601u/debugfs.c:92:6: error: format > > > > specifies type 'unsigned char' but the argument has type 'int' > > > > [-Werror,-Wformat] dev->ee->reg.start + dev->ee->reg.num - 1); [] > > I suggest s/%hh/%/ for all the uses here, not just this one. > > I also contemplated this change but I think it might be a bit out of > scope for https://github.com/ClangBuiltLinux/linux/issues/378 -- What > do you think? It would not be out of scope. > It could be argued that every single instance of %hh[dux] should be > replaced with %[dux]. All the vsprintf ones, but not the scanf ones.
Justin Stitt <justinstitt@google.com> wrote: > When building with Clang we encounter this warning: > | drivers/net/wireless/mediatek/mt7601u/debugfs.c:92:6: error: format > | specifies type 'unsigned char' but the argument has type 'int' > | [-Werror,-Wformat] dev->ee->reg.start + dev->ee->reg.num - 1); > > The format specifier used is `%hhu` which describes a u8. Both > `dev->ee->reg.start` and `.num` are u8 as well. However, the expression > as a whole is promoted to an int as you cannot get smaller-than-int from > addition. Therefore, to fix the warning, use the promoted-to-type's > format specifier -- in this case `%d`. > > example: > ``` > uint8_t a = 4, b = 7; > int size = sizeof(a + b - 1); > printf("%d\n", size); > // output: 4 > ``` > > See more: > (https://wiki.sei.cmu.edu/confluence/display/c/INT02-C.+Understand+integer+conversion+rules) > "Integer types smaller than int are promoted when an operation is > performed on them. If all values of the original type can be represented > as an int, the value of the smaller type is converted to an int; > otherwise, it is converted to an unsigned int." > > Signed-off-by: Justin Stitt <justinstitt@google.com> > Acked-by: Jakub Kicinski <kubakici@wp.pl> Patch applied to wireless-next.git, thanks. 68204a696505 wifi: mt7601u: fix clang -Wformat warning
diff --git a/drivers/net/wireless/mediatek/mt7601u/debugfs.c b/drivers/net/wireless/mediatek/mt7601u/debugfs.c index 20669eacb66e..230b0e1061a7 100644 --- a/drivers/net/wireless/mediatek/mt7601u/debugfs.c +++ b/drivers/net/wireless/mediatek/mt7601u/debugfs.c @@ -88,7 +88,7 @@ mt7601u_eeprom_param_show(struct seq_file *file, void *data) dev->ee->rssi_offset[0], dev->ee->rssi_offset[1]); seq_printf(file, "Reference temp: %hhx\n", dev->ee->ref_temp); seq_printf(file, "LNA gain: %hhx\n", dev->ee->lna_gain); - seq_printf(file, "Reg channels: %hhu-%hhu\n", dev->ee->reg.start, + seq_printf(file, "Reg channels: %hhu-%d\n", dev->ee->reg.start, dev->ee->reg.start + dev->ee->reg.num - 1); seq_puts(file, "Per rate power:\n");
When building with Clang we encounter this warning: | drivers/net/wireless/mediatek/mt7601u/debugfs.c:92:6: error: format | specifies type 'unsigned char' but the argument has type 'int' | [-Werror,-Wformat] dev->ee->reg.start + dev->ee->reg.num - 1); The format specifier used is `%hhu` which describes a u8. Both `dev->ee->reg.start` and `.num` are u8 as well. However, the expression as a whole is promoted to an int as you cannot get smaller-than-int from addition. Therefore, to fix the warning, use the promoted-to-type's format specifier -- in this case `%d`. example: ``` uint8_t a = 4, b = 7; int size = sizeof(a + b - 1); printf("%d\n", size); // output: 4 ``` See more: (https://wiki.sei.cmu.edu/confluence/display/c/INT02-C.+Understand+integer+conversion+rules) "Integer types smaller than int are promoted when an operation is performed on them. If all values of the original type can be represented as an int, the value of the smaller type is converted to an int; otherwise, it is converted to an unsigned int." Signed-off-by: Justin Stitt <justinstitt@google.com> --- Note: This patch silences the -Wformat warning for this file (which is the goal) but in reality all instances of `%hh[dux]` should be converted to `%[dux]` for this file and probably every file. That's a bit larger scope than the goal of enabling -Wformat for Clang builds, though. drivers/net/wireless/mediatek/mt7601u/debugfs.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)