| Message ID | 1340985513-15655-1-git-send-email-andriy.shevchenko@linux.intel.com (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Headers | show |
On Fri, 2012-06-29 at 18:58 +0300, Andy Shevchenko wrote: > There are many places in the kernel where the drivers print small buffers as a > hex string. This patch adds a support of the variable width buffer to print it > as a hex string with a delimiter. The idea came from Pavel Roskin here: > http://www.digipedia.pl/usenet/thread/18835/17449/ Seems sensible, but one stack caveat below > diff --git a/lib/vsprintf.c b/lib/vsprintf.c [] > @@ -655,11 +655,12 @@ char *resource_string(char *buf, char *end, struct resource *res, > } > > static noinline_for_stack > -char *mac_address_string(char *buf, char *end, u8 *addr, > - struct printf_spec spec, const char *fmt) > +char *hex_string(char *buf, char *end, u8 *addr, struct printf_spec spec, > + const char *fmt) > { > - char mac_addr[sizeof("xx:xx:xx:xx:xx:xx")]; > - char *p = mac_addr; > + char hex_str[64*3]; /* support up to 64 bytes to print */ Might be too much stack though. -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Fri, 29 Jun 2012 09:08:06 -0700 Joe Perches <joe@perches.com> wrote: > > diff --git a/lib/vsprintf.c b/lib/vsprintf.c > [] > > @@ -655,11 +655,12 @@ char *resource_string(char *buf, char *end, struct resource *res, > > } > > > > static noinline_for_stack > > -char *mac_address_string(char *buf, char *end, u8 *addr, > > - struct printf_spec spec, const char *fmt) > > +char *hex_string(char *buf, char *end, u8 *addr, struct printf_spec spec, > > + const char *fmt) > > { > > - char mac_addr[sizeof("xx:xx:xx:xx:xx:xx")]; > > - char *p = mac_addr; > > + char hex_str[64*3]; /* support up to 64 bytes to print */ > > Might be too much stack though. Yes, it's a bit marginal, as this new capability might be used in debug or crash situations where we're deep into the stack. The average case could be improved by using alloca()-style allocation. Documentation/printk-formats.txt would need to be updated please. Also the big comment over vsnprintf(). -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Fri, 2012-06-29 at 16:26 -0700, Andrew Morton wrote: > On Fri, 29 Jun 2012 09:08:06 -0700 > Joe Perches <joe@perches.com> wrote: > > > > diff --git a/lib/vsprintf.c b/lib/vsprintf.c > > [] > > > @@ -655,11 +655,12 @@ char *resource_string(char *buf, char *end, struct resource *res, > > > } > > > > > > static noinline_for_stack > > > -char *mac_address_string(char *buf, char *end, u8 *addr, > > > - struct printf_spec spec, const char *fmt) > > > +char *hex_string(char *buf, char *end, u8 *addr, struct printf_spec spec, > > > + const char *fmt) > > > { > > > - char mac_addr[sizeof("xx:xx:xx:xx:xx:xx")]; > > > - char *p = mac_addr; > > > + char hex_str[64*3]; /* support up to 64 bytes to print */ > > > > Might be too much stack though. > > Yes, it's a bit marginal, as this new capability might be used in debug > or crash situations where we're deep into the stack. The average case > could be improved by using alloca()-style allocation. Or maybe support larger sizes with a smaller stack buffer and a while loop. -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sat, Jun 30, 2012 at 5:48 PM, Joe Perches <joe@perches.com> wrote: > On Fri, 2012-06-29 at 16:26 -0700, Andrew Morton wrote: >> On Fri, 29 Jun 2012 09:08:06 -0700 >> Joe Perches <joe@perches.com> wrote: >> >> > > diff --git a/lib/vsprintf.c b/lib/vsprintf.c >> > [] >> > > @@ -655,11 +655,12 @@ char *resource_string(char *buf, char *end, struct resource *res, >> > > } >> > > >> > > static noinline_for_stack >> > > -char *mac_address_string(char *buf, char *end, u8 *addr, >> > > - struct printf_spec spec, const char *fmt) >> > > +char *hex_string(char *buf, char *end, u8 *addr, struct printf_spec spec, >> > > + const char *fmt) >> > > { >> > > - char mac_addr[sizeof("xx:xx:xx:xx:xx:xx")]; >> > > - char *p = mac_addr; >> > > + char hex_str[64*3]; /* support up to 64 bytes to print */ >> > >> > Might be too much stack though. >> >> Yes, it's a bit marginal, as this new capability might be used in debug >> or crash situations where we're deep into the stack. The average case >> could be improved by using alloca()-style allocation. > > Or maybe support larger sizes with a smaller > stack buffer and a while loop. What do you think about mixed approach? Let's say we would use buffer on stack for 8 bytes or less, and allocated buffer in case of larger input. It allows to keep implementation simple.
On Mon, 2012-07-02 at 20:32 +0300, Andy Shevchenko wrote: > On Sat, Jun 30, 2012 at 5:48 PM, Joe Perches <joe@perches.com> wrote: > > On Fri, 2012-06-29 at 16:26 -0700, Andrew Morton wrote: > >> On Fri, 29 Jun 2012 09:08:06 -0700 > >> Joe Perches <joe@perches.com> wrote: > >> > >> > > diff --git a/lib/vsprintf.c b/lib/vsprintf.c > >> > [] > >> > > @@ -655,11 +655,12 @@ char *resource_string(char *buf, char *end, struct resource *res, > >> > > } > >> > > > >> > > static noinline_for_stack > >> > > -char *mac_address_string(char *buf, char *end, u8 *addr, > >> > > - struct printf_spec spec, const char *fmt) > >> > > +char *hex_string(char *buf, char *end, u8 *addr, struct printf_spec spec, > >> > > + const char *fmt) > >> > > { > >> > > - char mac_addr[sizeof("xx:xx:xx:xx:xx:xx")]; > >> > > - char *p = mac_addr; > >> > > + char hex_str[64*3]; /* support up to 64 bytes to print */ > >> > > >> > Might be too much stack though. > >> > >> Yes, it's a bit marginal, as this new capability might be used in debug > >> or crash situations where we're deep into the stack. The average case > >> could be improved by using alloca()-style allocation. > > > > Or maybe support larger sizes with a smaller > > stack buffer and a while loop. > > What do you think about mixed approach? Let's say we would use buffer > on stack for 8 bytes or less, and allocated buffer in case of larger > input. It allows to keep implementation simple. > I think the while loop is simplest. I'll code something up tomorrow unless you get to it first. -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/lib/vsprintf.c b/lib/vsprintf.c index a0b5f15..1645d7e 100644 --- a/lib/vsprintf.c +++ b/lib/vsprintf.c @@ -655,11 +655,12 @@ char *resource_string(char *buf, char *end, struct resource *res, } static noinline_for_stack -char *mac_address_string(char *buf, char *end, u8 *addr, - struct printf_spec spec, const char *fmt) +char *hex_string(char *buf, char *end, u8 *addr, struct printf_spec spec, + const char *fmt) { - char mac_addr[sizeof("xx:xx:xx:xx:xx:xx")]; - char *p = mac_addr; + char hex_str[64*3]; /* support up to 64 bytes to print */ + int len = 6; /* default length is 6 bytes */ + char *p = hex_str; int i; char separator; bool reversed = false; @@ -678,18 +679,21 @@ char *mac_address_string(char *buf, char *end, u8 *addr, break; } - for (i = 0; i < 6; i++) { + if (spec.field_width > 0) + len = min_t(int, spec.field_width, 64); + + for (i = 0; i < len; i++) { if (reversed) - p = hex_byte_pack(p, addr[5 - i]); + p = hex_byte_pack(p, addr[len - 1 - i]); else p = hex_byte_pack(p, addr[i]); - if (fmt[0] == 'M' && i != 5) + if (fmt[0] == 'M' && i != len - 1) *p++ = separator; } *p = '\0'; - return string(buf, end, mac_addr, spec); + return string(buf, end, hex_str, spec); } static noinline_for_stack @@ -1011,7 +1015,7 @@ char *pointer(const char *fmt, char *buf, char *end, void *ptr, case 'm': /* Contiguous: 000102030405 */ /* [mM]F (FDDI) */ /* [mM]R (Reverse order; Bluetooth) */ - return mac_address_string(buf, end, ptr, spec, fmt); + return hex_string(buf, end, ptr, spec, fmt); case 'I': /* Formatted IP supported * 4: 1.2.3.4 * 6: 0001:0203:...:0708
There are many places in the kernel where the drivers print small buffers as a hex string. This patch adds a support of the variable width buffer to print it as a hex string with a delimiter. The idea came from Pavel Roskin here: http://www.digipedia.pl/usenet/thread/18835/17449/ Sample output of pr_info("buf[%d:%d] %*pM\n", from, len, len, &buf[from]); could be like this: [ 0.726130] buf[51:8] e8:16:b6:ef:e3:74:45:6e [ 0.750736] buf[59:15] 31:81:b8:3f:35:49:06:ae:df:32:06:05:4a:af:55 [ 0.757602] buf[17:5] ac:16:d5:2c:ef Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> --- lib/vsprintf.c | 22 +++++++++++++--------- 1 file changed, 13 insertions(+), 9 deletions(-)