a7bc9ac8f
Denis Vlasenko
2007-09-28
Subject
hdparm: reduce possibility of numeric overflow in -T.
Binary
No binary (build failed)
Architecture & Compiler Flags
—
Build Log
(log yok)
Commit Diff GitHub →
miscutils/hdparm.c+14 -6
| @@ -1375,10 +1375,13 @@ static void print_timing(unsigned long m, unsigned elapsed_us) | |
| printf("%5lu MB in %u.%02u seconds = %lu kB/s\n", | |
| m, sec, hs, | |
| /* Trying to not overflow 32-bit arith in m * CONST | |
| - | * by keeping CONST not so big. + 1 prevents div-by-0. */ |
| - | (m * (1024 * 1000000 / (64*1024))) / (elapsed_us / (64*1024) + 1) |
| - | // ~= (m * 1024 * 1000000) / elapsed_ms |
| - | // = (m * 1024) / (elapsed_ms / 1000000) |
| + | * by keeping CONST not so big. But elapsed_us / CONST2 |
| + | * also should not introduce big errors. Currently, |
| + | * 16000us is ~1.6% of 1 second. |
| + | * "+ 1" prevents div-by-0. */ |
| + | (m * (1024 * 1000000 / (16*1024))) / (elapsed_us / (16*1024) + 1) |
| + | // ~= (m * 1024 * 1000000) / elapsed_us |
| + | // = (m * 1024) / (elapsed_us / 1000000) |
| // = kb / elapsed_sec | |
| ); | |
| } | |
| @@ -1418,6 +1421,11 @@ static void do_time(int cache /*,int fd*/) | |
| /* Now do the timing */ | |
| iterations = 0; | |
| + | /* Max time to run (small for cache, avoids getting |
| + | * huge total_MB which can overlow on print_timing) */ |
| + | elapsed2 = 510000; /* cache */ |
| + | if (!cache) |
| + | elapsed2 = 3000000; /* not cache */ |
| start = monotonic_us(); | |
| do { | |
| if (cache) | |
| @@ -1425,11 +1433,11 @@ static void do_time(int cache /*,int fd*/) | |
| read_big_block(buf); | |
| elapsed = (unsigned)monotonic_us() - start; | |
| ++iterations; | |
| - | } while (elapsed < 3000000 && iterations < max_iterations); |
| + | } while (elapsed < elapsed2 && iterations < max_iterations); |
| total_MB = (unsigned long)iterations * TIMING_BUF_MB; | |
| //printf(" elapsed:%u iterations:%u ", elapsed, iterations); | |
| if (cache) { | |
| - | /* Now remove the lseek() and monotonic_us() overheads |
| + | /* Cache: remove lseek() and monotonic_us() overheads |
| * from elapsed */ | |
| start = monotonic_us(); | |
| do { | |