| Message ID | 1309376974-12943-1-git-send-email-sjg@chromium.org |
|---|---|
| State | Accepted, archived |
| Headers | show |
On Wednesday, June 29, 2011 15:49:34 Simon Glass wrote:
> - Use separate __assert_fail() function to reduce memory footprint
do we really care about that when people are using #define DEBUG ? i'd say
this is an unnecessary indirection ...
-mike
Dear Mike Frysinger, In message <201106291643.32477.vapier@gentoo.org> you wrote: > > do we really care about that when people are using #define DEBUG ? i'd say > this is an unnecessary indirection ... Yes, we do care about the memory footprint - no matter whether with or without DEBUG. Best regards, Wolfgang Denk
On Wednesday, June 29, 2011 17:01:23 Wolfgang Denk wrote: > Mike Frysinger wrote: > > do we really care about that when people are using #define DEBUG ? i'd > > say this is an unnecessary indirection ... > > Yes, we do care about the memory footprint - no matter whether with or > without DEBUG. there's caring about it, and there's nitpicking over ~10 bytes per call site. -mike
Did this go anywhere? I will resend... - Simon On Wed, Jun 29, 2011 at 2:54 PM, Mike Frysinger <vapier@gentoo.org> wrote: > On Wednesday, June 29, 2011 17:01:23 Wolfgang Denk wrote: >> Mike Frysinger wrote: >> > do we really care about that when people are using #define DEBUG ? i'd >> > say this is an unnecessary indirection ... >> >> Yes, we do care about the memory footprint - no matter whether with or >> without DEBUG. > > there's caring about it, and there's nitpicking over ~10 bytes per call site. > -mike >
On Wednesday, September 07, 2011 18:29:16 Simon Glass wrote: > On Wed, Jun 29, 2011 at 2:54 PM, Mike Frysinger <vapier@gentoo.org> wrote: > > On Wednesday, June 29, 2011 17:01:23 Wolfgang Denk wrote: > >> Mike Frysinger wrote: > >> > do we really care about that when people are using #define DEBUG ? > >> > i'd say this is an unnecessary indirection ... > >> > >> Yes, we do care about the memory footprint - no matter whether with or > >> without DEBUG. > > > > there's caring about it, and there's nitpicking over ~10 bytes per call > > site. > > Did this go anywhere? I will resend... i dont care enough to fight wolfgang over it -mike
Hi Mike / Wolfgang, On Thu, Sep 8, 2011 at 5:34 PM, Mike Frysinger <vapier@gentoo.org> wrote: > On Wednesday, September 07, 2011 18:29:16 Simon Glass wrote: >> On Wed, Jun 29, 2011 at 2:54 PM, Mike Frysinger <vapier@gentoo.org> wrote: >> > On Wednesday, June 29, 2011 17:01:23 Wolfgang Denk wrote: >> >> Mike Frysinger wrote: >> >> > do we really care about that when people are using #define DEBUG ? >> >> > i'd say this is an unnecessary indirection ... >> >> >> >> Yes, we do care about the memory footprint - no matter whether with or >> >> without DEBUG. >> > >> > there's caring about it, and there's nitpicking over ~10 bytes per call >> > site. >> >> Did this go anywhere? I will resend... > > i dont care enough to fight wolfgang over it OK. Any other comments on this Wolfgang or is this finished? Regards, Simon > -mike >
Dear Simon Glass, In message <1309376974-12943-1-git-send-email-sjg@chromium.org> you wrote: > assert() is like BUG_ON() but compiles to nothing unless DEBUG is defined. > This is useful when a condition is an error but a board reset is unlikely > to fix it, so it is better to soldier on in hope. Assertion failures should > be caught during development/test. > > It turns out that assert() is defined separately in a few places in U-Boot > with various meanings. This patch cleans up some of these. > > Build errors exposed by this change (and defining DEBUG) are also fixed in > this patch. > > Signed-off-by: Simon Glass <sjg@chromium.org> > --- > Changes in v2: > - Changed macros so that all code is compiled even if DEBUG is disabled > > Changes in v3: > - Use panic() instead of printf() > - Use separate __assert_fail() function to reduce memory footprint > - Changed output format to match assert(3) > > common/dlmalloc.c | 7 ------- > include/common.h | 21 +++++++++++++++++++++ > include/malloc.h | 8 -------- > lib/qsort.c | 5 ----- > lib/vsprintf.c | 8 ++++++++ > 5 files changed, 29 insertions(+), 20 deletions(-) Applied, thanks. Best regards, Wolfgang Denk
diff --git a/common/dlmalloc.c b/common/dlmalloc.c index e9bab09..f2080c6 100644 --- a/common/dlmalloc.c +++ b/common/dlmalloc.c @@ -286,13 +286,6 @@ extern "C" { */ -#ifdef DEBUG -#include <assert.h> -#else -#define assert(x) ((void)0) -#endif - - /* INTERNAL_SIZE_T is the word-size used for internal bookkeeping of chunk sizes. On a 64-bit machine, you can reduce malloc diff --git a/include/common.h b/include/common.h index 1e21b7a..0956b93 100644 --- a/include/common.h +++ b/include/common.h @@ -124,6 +124,27 @@ typedef volatile unsigned char vu_char; #define debugX(level,fmt,args...) #endif /* DEBUG */ +#ifdef DEBUG +# define _DEBUG 1 +#else +# define _DEBUG 0 +#endif + +/* + * An assertion is run-time check done in debug mode only. If DEBUG is not + * defined then it is skipped. If DEBUG is defined and the assertion fails, + * then it calls panic*( which may or may not reset/halt U-Boot (see + * CONFIG_PANIC_HANG), It is hoped that all failing assertions are found + * before release, and after release it is hoped that they don't matter. But + * in any case these failing assertions cannot be fixed with a reset (which + * may just do the same assertion again). + */ +void __assert_fail(const char *assertion, const char *file, unsigned line, + const char *function); +#define assert(x) \ + ({ if (!(x) && _DEBUG) \ + __assert_fail(#x, __FILE__, __LINE__, __func__); }) + #define error(fmt, args...) do { \ printf("ERROR: " fmt "\nat %s:%d/%s()\n", \ ##args, __FILE__, __LINE__, __func__); \ diff --git a/include/malloc.h b/include/malloc.h index 3e145ad..ecf3c67 100644 --- a/include/malloc.h +++ b/include/malloc.h @@ -285,14 +285,6 @@ extern "C" { */ -#ifdef DEBUG -/* #include <assert.h> */ -#define assert(x) ((void)0) -#else -#define assert(x) ((void)0) -#endif - - /* INTERNAL_SIZE_T is the word-size used for internal bookkeeping of chunk sizes. On a 64-bit machine, you can reduce malloc diff --git a/lib/qsort.c b/lib/qsort.c index 1cc0d31..86c392c 100644 --- a/lib/qsort.c +++ b/lib/qsort.c @@ -17,11 +17,6 @@ #include <linux/types.h> #include <exports.h> -#if 0 -#include <assert.h> -#else -#define assert(arg) -#endif void qsort(void *base, size_t nel, diff --git a/lib/vsprintf.c b/lib/vsprintf.c index 3b924ec..1f01b70 100644 --- a/lib/vsprintf.c +++ b/lib/vsprintf.c @@ -728,3 +728,11 @@ void panic(const char *fmt, ...) do_reset (NULL, 0, 0, NULL); #endif } + +void __assert_fail(const char *assertion, const char *file, unsigned line, + const char *function) +{ + /* This will not return */ + panic("%s:%u: %s: Assertion `%s' failed.", file, line, function, + assertion); +}
assert() is like BUG_ON() but compiles to nothing unless DEBUG is defined. This is useful when a condition is an error but a board reset is unlikely to fix it, so it is better to soldier on in hope. Assertion failures should be caught during development/test. It turns out that assert() is defined separately in a few places in U-Boot with various meanings. This patch cleans up some of these. Build errors exposed by this change (and defining DEBUG) are also fixed in this patch. Signed-off-by: Simon Glass <sjg@chromium.org> --- Changes in v2: - Changed macros so that all code is compiled even if DEBUG is disabled Changes in v3: - Use panic() instead of printf() - Use separate __assert_fail() function to reduce memory footprint - Changed output format to match assert(3) common/dlmalloc.c | 7 ------- include/common.h | 21 +++++++++++++++++++++ include/malloc.h | 8 -------- lib/qsort.c | 5 ----- lib/vsprintf.c | 8 ++++++++ 5 files changed, 29 insertions(+), 20 deletions(-)