module: fix sprintf format specifier in param_get_byte()
PATCH 1/1 module: fix sprintf format specifier in param_get_byte()
In param_get_byte(), to which the macro STANDARD_PARAM_DEF(byte, ...) expands, "%c" is used to print an unsigned char. So it gets printed as a character what is not intended here. Use "%hhu" instead.
--- kernel/params.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/params.c +1 -1
@@ -252,7 +252,7 @@ int parse_args(const char *doing, EXPORT_SYMBOL(param_ops_##name) -STANDARD_PARAM_DEF(byte, unsigned char, "%c", unsigned long, strict_strtoul); +STANDARD_PARAM_DEF(byte, unsigned char, "%hhu", unsigned long, strict_strtoul);STANDARD_PARAM_DEF (short, short,"%hi" , long, strict_strtol); STANDARD_PARAM_DEF(ushort, unsigned short, "%hu", unsigned long, strict_strtoul);STANDARD_PARAM_DEF (int, int,"%i" , long, strict_strtol);
Signature · 1 line
1.8.3.1
Quoted 14 lines · Christoph Jaeger
In
param_get_byte(), to which the macroSTANDARD_PARAM_DEF(byte, ...)expands, "%c" is used to print an unsigned char. So it gets printed as a character what is not intended here. Use "%hhu" instead.Signed-off-by Christoph Jaeger <[email protected]>--- kernel/params.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)kernel/params.c +0
-0
kernel/params.c int parse_args(const char *doing, · line 257
Quoted 7 lines
kernel/params.c +1
-1@@ -252,7 +252,7 @@ int parse_args(const char *doing, EXPORT_SYMBOL(param_ops_##name) -STANDARD_PARAM_DEF(byte, unsigned char, "%c", unsigned long, strict_strtoul); +STANDARD_PARAM_DEF(byte, unsigned char, "%hhu", unsigned long, strict_strtoul);STANDARD_PARAM_DEF (short, short,"%hi" , long, strict_strtol); STANDARD_PARAM_DEF(ushort, unsigned short, "%hu", unsigned long, strict_strtoul);
I don't see the point of using "%hh[ui]" and "%h[ui]". These are promoted to int/unsigned int for the sprintf anyway. I'd just use %d and %u instead.
Quoted 5 lines · Christoph Jaeger
In
param_get_byte(), to which the macroSTANDARD_PARAM_DEF(byte, ...)expands, "%c" is used to print an unsigned char. So it gets printed as a character what is not intended here. Use "%hhu" instead.Signed-off-by Christoph Jaeger <[email protected]>
Nice patch. Unfortunately, there are several users of this already:
drivers/net/wireless/cw1200/main.c:50:module_param_array_named(macaddr, cw1200_mac_template, byte, NULL, S_IRUGO); drivers/ntb/ntb_transport.c:68:module_param(max_num_clients, byte, 0644); drivers/scsi/lpfc/lpfc_attr.c:4207:module_param(lpfc_prot_guard, byte, S_IRUGO); drivers/usb/atm/speedtch.c:117:module_param(ModemMode, byte, S_IRUGO | S_IWUSR); drivers/usb/atm/speedtch.c:121:module_param_array(ModemOption, byte, &num_ModemOption, S_IRUGO);
drivers/usb/gadget/g_ffs.c:95:module_param_named(bDeviceClass, gfs_dev_desc.bDeviceClass, byte, 0644); drivers/usb/gadget/g_ffs.c:97:module_param_named(bDeviceSubClass, gfs_dev_desc.bDeviceSubClass, byte, 0644); drivers/usb/gadget/g_ffs.c:99:module_param_named(bDeviceProtocol, gfs_dev_desc.bDeviceProtocol, byte, 0644);
I have CC'd all the authors, to see if changing the results of reading /sys/module/<modname>/parameters/<xxx> from a literal char to a number will harm them.
Thanks,
Rusty.
Quoted 19 lines
--- kernel/params.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)kernel/params.c +1
-1@@ -252,7 +252,7 @@ int parse_args(const char *doing, EXPORT_SYMBOL(param_ops_##name) -STANDARD_PARAM_DEF(byte, unsigned char, "%c", unsigned long, strict_strtoul); +STANDARD_PARAM_DEF(byte, unsigned char, "%hhu", unsigned long, strict_strtoul);STANDARD_PARAM_DEF (short, short,"%hi" , long, strict_strtol); STANDARD_PARAM_DEF(ushort, unsigned short, "%hu", unsigned long, strict_strtoul);STANDARD_PARAM_DEF (int, int,"%i" , long, strict_strtol);Signature · 1 line
1.8.3.1
Michal Nazarewicz
Acked-byQuoted 6 lines · Rusty Russell
Quoted 5 lines · Christoph Jaeger
In
param_get_byte(), to which the macroSTANDARD_PARAM_DEF(byte, ...)expands, "%c" is used to print an unsigned char. So it gets printed as a character what is not intended here. Use "%hhu" instead.Signed-off-by Christoph Jaeger <[email protected]>
for g_ffs.c:
Quoted 3 lines
drivers/usb/gadget/g_ffs.c:95:module_param_named(bDeviceClass, gfs_dev_desc.bDeviceClass, byte, 0644); drivers/usb/gadget/g_ffs.c:97:module_param_named(bDeviceSubClass, gfs_dev_desc.bDeviceSubClass, byte, 0644); drivers/usb/gadget/g_ffs.c:99:module_param_named(bDeviceProtocol, gfs_dev_desc.bDeviceProtocol, byte, 0644);
I don't think it breaks anything for g_ffs since those properties are most likely write-only and I don't expect many people reading them.
Quoted 12 lines
Quoted 12 lines
--- kernel/params.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)kernel/params.c +0
-0@@ -252,7 +252,7 @@ int parse_args(const char *doing, EXPORT_SYMBOL(param_ops_##name)
Quoted 4 lines
kernel/params.c int parse_args(const char *doing, · line 257
Quoted 4 lines
kernel/params.c +1
-1-STANDARD_PARAM_DEF(byte, unsigned char, "%c", unsigned long, strict_strtoul); +STANDARD_PARAM_DEF(byte, unsigned char, "%hhu", unsigned long, strict_strtoul);STANDARD_PARAM_DEF (short, short,"%hi" , long, strict_strtol); STANDARD_PARAM_DEF(ushort, unsigned short, "%hu", unsigned long, strict_strtoul);
Actually, are those “h” specifiers even necessary? I'm fairly certain that “%u” for byte, “%i” for short, and “%u” for ushort would work just fine since the argument gets promoted to (unsigned) int anyway and indeed vsnprintf reads an int for all those types.
Quoted 1 line
Quoted 1 line
kernel/params.c +0
-0STANDARD_PARAM_DEF (int, int,"%i" , long, strict_strtol);
Signature · 4 lines
Best regards, _ _ .o. | Liege of Serenely Enlightened Majesty of o' \,=./ `o ..o | Computer Science, Michał “mina86” Nazarewicz (o o) ooo +----<email/xmpp: [email protected]>--------------ooO--(_)--Ooo--
Jon Mason
Acked-byQuoted 11 lines · Rusty Russell
Quoted 5 lines · Christoph Jaeger
In
param_get_byte(), to which the macroSTANDARD_PARAM_DEF(byte, ...)expands, "%c" is used to print an unsigned char. So it gets printed as a character what is not intended here. Use "%hhu" instead.Signed-off-by Christoph Jaeger <[email protected]>Nice patch. Unfortunately, there are several users of this already:
drivers/net/wireless/cw1200/main.c:50:module_param_array_named(macaddr,
cw1200_mac_template, byte, NULL,S_IRUGO); drivers/ntb/ntb_transport.c:68:module_param(max_num_clients, byte, 0644);
This shouldn't affect NTB negatively.
Quoted 33 lines
drivers/scsi/lpfc/lpfc_attr.c:4207:module_param(lpfc_prot_guard, byte,
S_IRUGO); drivers/usb/atm/speedtch.c:117:module_param(ModemMode, byte,S_IRUGO|S_IWUSR); drivers/usb/atm/speedtch.c:121:module_param_array(ModemOption, byte,&num_ModemOption,S_IRUGO);drivers/usb/gadget/g_ffs.c:95:module_param_named(bDeviceClass, gfs_dev_desc.bDeviceClass, byte, 0644); drivers/usb/gadget/g_ffs.c:97:module_param_named(bDeviceSubClass, gfs_dev_desc.bDeviceSubClass, byte, 0644); drivers/usb/gadget/g_ffs.c:99:module_param_named(bDeviceProtocol, gfs_dev_desc.bDeviceProtocol, byte, 0644);I have CC'd all the authors, to see if changing the results of reading /sys/module/<modname>/parameters/<xxx> from a literal char to a number will harm them.
Thanks,
Rusty.Quoted 19 lines
--- kernel/params.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)kernel/params.c +1
-1@@ -252,7 +252,7 @@ int parse_args(const char *doing, EXPORT_SYMBOL(param_ops_##name) -STANDARD_PARAM_DEF(byte, unsigned char, "%c", unsigned long, strict_strtoul); +STANDARD_PARAM_DEF(byte, unsigned char, "%hhu", unsigned long, strict_strtoul);STANDARD_PARAM_DEF (short, short,"%hi" , long, strict_strtol); STANDARD_PARAM_DEF(ushort, unsigned short, "%hu", unsigned long, strict_strtoul);STANDARD_PARAM_DEF (int, int,"%i" , long, strict_strtol);Signature · 1 line
1.8.3.1