writeback: use |1 instead of +1 to protect against div by zero
Subsystems (2)
PATCH 1/1 writeback: use |1 instead of +1 to protect against div by zero
mm/page-writeback.c has several places where 1 is added to the divisor to prevent division by zero exceptions; however, if the original divisor is equivalent to -1, adding 1 leads to division by zero.
There are three places where +1 is used for this purpose - one in pos_ratio_polynom() and two in bdi_position_ratio(). The second one in bdi_position_ratio() actually triggered div-by-zero oops on a machine running a 3.10 kernel. The divisor is
x_intercept - bdi_setpoint + 1 == span + 1
span is confirmed to be (u32)-1. It isn't clear how it ended up that but it could be from write bandwidth calculation underflow fixed by c72efb658f7c ("writeback: fix possible underflow in write bandwidth calculation").
At any rate, +1 isn't a proper protection against div-by-zero. This patch converts all +1 protections to |1. Note that bdi_update_dirty_ratelimit() was already using |1 before this patch.
--- mm/page-writeback.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
mm/page-writeback.c +3 -3
@@ -580,7 +580,7 @@ static long long pos_ratio_polynom(unsigned long setpoint,long x; x =div64_s64 (((s64 )setpoint - (s64 )dirty) <<RATELIMIT_CALC_SHIFT , - limit - setpoint +1 ); + (limit - setpoint) |1 ); pos_ratio = x; pos_ratio = pos_ratio * x >>RATELIMIT_CALC_SHIFT ; pos_ratio = pos_ratio * x >>RATELIMIT_CALC_SHIFT ;
@@ -807,7 +807,7 @@ static unsigned long bdi_position_ratio(struct backing_dev_info *bdi,* scale global setpoint to bdi's: * bdi_setpoint = setpoint * bdi_thresh / thresh */ - x =div_u64 ((u64 )bdi_thresh <<16 , thresh +1 ); + x =div_u64 ((u64 )bdi_thresh <<16 , thresh |1 ); bdi_setpoint = setpoint * (u64 )x >>16 ;/* * Use span=(8*write_bw) in single bdi case as indicated by
@@ -822,7 +822,7 @@ static unsigned long bdi_position_ratio(struct backing_dev_info *bdi,if (bdi_dirty < x_intercept - span /4 ) { pos_ratio =div64_u64 (pos_ratio * (x_intercept - bdi_dirty), - x_intercept - bdi_setpoint +1 ); + (x_intercept - bdi_setpoint) |1 ); }else pos_ratio /=4 ;
Quoted 11 lines · Tejun Heo
mm/page-writeback.chas several places where 1 is added to the divisor to prevent division by zero exceptions; however, if the original divisor is equivalent to -1, adding 1 leads to division by zero.There are three places where +1 is used for this purpose - one in
pos_ratio_polynom()and two inbdi_position_ratio(). The second one inbdi_position_ratio()actually triggered div-by-zero oops on a machine running a 3.10 kernel. The divisor is
x_intercept-bdi_setpoint+ 1 == span + 1
Quoted 8 lines
span is confirmed to be (u32)-1. It isn't clear how it ended up that but it could be from write bandwidth calculation underflow fixed by
c72efb658f7c("writeback: fix possible underflow in write bandwidth calculation").At any rate, +1 isn't a proper protection against div-by-zero. This patch converts all +1 protections to |1. Note that
bdi_update_dirty_ratelimit()was already using |1 before this patch.
So IMHO the only proper protection against division by zero is
if (unlikely(divisor == 0)) return some_value; else do division
But we took a shortcut with +1 or |1 in those calculations since that
should be OK given the *expected* numbers we deal with. Once numbers get out of expected range, both shortcuts have issues, just different ones (sure you avoid division by zero with |1 but the absurd values you'll be getting will have interesting effect on the system anyway). So I'm OK with changing +1 to |1 to avoid oops and make code consistent, I just wanted to comment on the fact that the change is for avoiding oops, not about getting better values.
Otherwise feel free to add:
Honza
Quoted 31 lines
mm/page-writeback.c +3
-3@@ -580,7 +580,7 @@ static long long pos_ratio_polynom(unsigned long setpoint,long x; x =div64_s64 (((s64 )setpoint - (s64 )dirty) <<RATELIMIT_CALC_SHIFT , - limit - setpoint +1 ); + (limit - setpoint) |1 ); pos_ratio = x; pos_ratio = pos_ratio * x >>RATELIMIT_CALC_SHIFT ; pos_ratio = pos_ratio * x >>RATELIMIT_CALC_SHIFT ;@@ -807,7 +807,7 @@ static unsigned long bdi_position_ratio(struct backing_dev_info *bdi,* scale global setpoint to bdi's: * bdi_setpoint = setpoint * bdi_thresh / thresh */ - x =div_u64 ((u64 )bdi_thresh <<16 , thresh +1 ); + x =div_u64 ((u64 )bdi_thresh <<16 , thresh |1 ); bdi_setpoint = setpoint * (u64 )x >>16 ;/* * Use span=(8*write_bw) in single bdi case as indicated by @@ -822,7 +822,7 @@ static unsigned long bdi_position_ratio(struct backing_dev_info *bdi,if (bdi_dirty < x_intercept - span /4 ) { pos_ratio =div64_u64 (pos_ratio * (x_intercept - bdi_dirty), - x_intercept - bdi_setpoint +1 ); + (x_intercept - bdi_setpoint) |1 ); }else pos_ratio /=4 ;
Signature · 2 lines
Jan Kara <[email protected]> SUSE Labs, CR
Quoted 11 lines · Tejun Heo
mm/page-writeback.chas several places where 1 is added to the divisor to prevent division by zero exceptions; however, if the original divisor is equivalent to -1, adding 1 leads to division by zero.There are three places where +1 is used for this purpose - one in
pos_ratio_polynom()and two inbdi_position_ratio(). The second one inbdi_position_ratio()actually triggered div-by-zero oops on a machine running a 3.10 kernel. The divisor isx_intercept - bdi_setpoint + 1 == span + 1
Quoted 8 lines
span is confirmed to be (u32)-1. It isn't clear how it ended up that but it could be from write bandwidth calculation underflow fixed by
c72efb658f7c("writeback: fix possible underflow in write bandwidth calculation").At any rate, +1 isn't a proper protection against div-by-zero. This patch converts all +1 protections to |1. Note that
bdi_update_dirty_ratelimit()was already using |1 before this patch.
The |1 is a litle iffy imho, but I guess it gets the job done. Applied for 4.1.
Signature · 1 line
Jens Axboe