diff options
| author | Haiying Wang <Haiying.Wang@freescale.com> | 2013-05-17 18:23:44 (GMT) |
|---|---|---|
| committer | Fleming Andrew-AFLEMING <AFLEMING@freescale.com> | 2013-05-17 20:01:02 (GMT) |
| commit | f24fff6a6f14a9938a200170bd9a0f40c238d58c (patch) | |
| tree | 163dfc03908345e68e352028e2b2b8a641130599 /include | |
| parent | ad1bdc90739189c25881f12e21940f84398b8ef0 (diff) | |
| download | linux-fsl-qoriq-f24fff6a6f14a9938a200170bd9a0f40c238d58c.tar.xz | |
qman: use math64 instead of direct 64-bit division.
In 32-bit builds, the ROUNDING macro was breaking kernel assumptions when
fed with 64-bit parameters. This forces it to use the recommended
wrappers in <linux/math64.h> instead.
Signed-off-by: Geoff Thorpe <Geoff.Thorpe@freescale.com>
Change-Id: Id708cdad58593f38683112adf59984ffd3d763f7
Reviewed-on: http://git.am.freescale.net:8181/2580
Tested-by: Review Code-CDREVIEW <CDREVIEW@freescale.com>
Reviewed-by: Fleming Andrew-AFLEMING <AFLEMING@freescale.com>
Diffstat (limited to 'include')
| -rw-r--r-- | include/linux/fsl_qman.h | 11 |
1 files changed, 0 insertions, 11 deletions
diff --git a/include/linux/fsl_qman.h b/include/linux/fsl_qman.h index da2c67d3..673724a 100644 --- a/include/linux/fsl_qman.h +++ b/include/linux/fsl_qman.h @@ -2417,17 +2417,6 @@ struct qm_ceetm_lfq { qman_cb_mr ern; }; -/* Divide 'n' by 'd', rounding down if 'r' is negative, rounding up if - * it's positive, and rounding to the closest value if it's zero. NB, - * this macro assumes no particular type, so feed it with types that are - * appropriate for its use. NB, these arguments should not be expressions - * unless it is safe for them to be evaluated multiple times. Eg. do not - * pass in "some_value++" as a parameter to the macro! */ -#define ROUNDING(n, d, r) \ - (((r) < 0) ? ((n) / (d)) : \ - (((r) > 0) ? (((n) + (d) - 1) / (d)) : \ - (((n) + ((d) / 2)) / (d)))) - /** * qman_ceetm_bps2tokenrate - Given a desired rate 'bps' measured in bps * (ie. bits-per-second), compute the 'token_rate' fraction that best |