common/memsize.c: fix endless loop when saving
PATCH 1/1 common/memsize.c: fix endless loop when saving
When cnt reaches 0, any shift operation will keep cnt=0, so the condition cnt >= 0 doesn't make sense.
To fix this endless loop, we drop the useless condition and simply break the loop when cnt reaches 0.
--- common/memsize.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
common/memsize.c +2 -1
@@ -33,7 +33,7 @@ long get_ram_size(long *base, long maxsize)long size;int i =0 ; -for (cnt = (maxsize /sizeof (long )) >>1 ; cnt >=0 ; cnt >>=1 ) { +for (cnt = (maxsize /sizeof (long )) >>1 ; ; cnt >>=1 ) { addr = base + cnt;/* pointer arith! */ sync (); save[i] = *addr;
@@ -43,6 +43,7 @@ long get_ram_size(long *base, long maxsize) *addr = ~cnt; }else { *addr =0 ; +break ; } }
Signature · 1 line
1.9.2.459.g68773ac
PATCH 1/1 common/memsize.c: fix endless loop when saving
2016-02-09 19:41 GMT+02:00 Eddy Petri?or <[email protected]>:
Quoted 5 lines
When cnt reaches 0, any shift operation will keep cnt=0, so the condition
cnt >= 0doesn't make sense.To fix this endless loop, we drop the useless condition and simply break the loop when cnt reaches 0.
This should fix the endless issue introduced by the previous patch 8e7cba048baae68ee0916a8f52b4304277328d5e
Hannes, can you confirm that with this patch the reported size is correct and the boot is normal again?
I initially thought the size might be incorret due to some buggy HW I have (it has some unstable DDR settings, which lead to incorrect reports on my side initially).
Quoted 29 lines
Signed-off-by Eddy Petri?or <[email protected]>--- common/memsize.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)common/memsize.c +2
-2@@ -33,7 +33,7 @@ long get_ram_size(long *base, long maxsize)long size;int i =0 ; -for (cnt = (maxsize /sizeof (long )) >>1 ; cnt >=0 ; cnt >>=1 ) { +for (cnt = (maxsize /sizeof (long )) >>1 ; ; cnt >>=1 ) { addr = base + cnt;/* pointer arith! */ sync (); save[i] = *addr;@@ -43,6 +43,7 @@ long get_ram_size(long *base, long maxsize) *addr = ~cnt; }else { *addr =0 ; +break ; } } --1.9.2.459.g68773ac
Signature · 1 line
Eddy Petri?or
PATCH 1/1 common/memsize.c: fix endless loop when saving
Quoted 6 lines · Eddy Petri?or
2016-02-09 19:41 GMT+02:00 Eddy Petri?or <[email protected]>:
Quoted 5 lines
When cnt reaches 0, any shift operation will keep cnt=0, so the condition
cnt >= 0doesn't make sense.To fix this endless loop, we drop the useless condition and simply break the loop when cnt reaches 0.
Quoted 8 lines
This should fix the endless issue introduced by the previous patch
8e7cba048baae68ee0916a8f52b4304277328d5eHannes, can you confirm that with this patch the reported size is correct and the boot is normal again?
I initially thought the size might be incorret due to some buggy HW I have (it has some unstable DDR settings, which lead to incorrect reports on my side initially).
Hi Eddy,
the board does boot with this patch.
But the reported memsize is wrong.
1GiB instead 256MiB.
So we need some v2 to get rid.
best regards,
Hannes
PATCH 1/1 common/memsize.c: fix endless loop when saving
Pe 10 feb. 2016 8:26 a.m., "Hannes Schmelzer" <[email protected]> a scris:
Quoted 11 lines · Eddy Petri?or
Quoted 8 lines
2016-02-09 19:41 GMT+02:00 Eddy Petri?or <[email protected]>:
Hannes, can you confirm that with this patch the reported size is correct and the boot is normal again?
I initially thought the size might be incorret due to some buggy HW I have (it has some unstable DDR settings, which lead to incorrect reports on my side initially).
Quoted 6 lines
Hi Eddy,
the board does boot with this patch.
But the reported memsize is wrong.
1GiB instead 256MiB.So we need some v2 to get rid.
Taking into account the first version was reverted, would you be willing to test a reworked patch targeting v2016.05, (when I have it ready)?
Quoted 4 lines
best regards,
Hannes
PATCH 1/1 common/memsize.c: fix endless loop when saving
Quoted 21 lines · Eddy Petri?or
Pe 10 feb. 2016 8:26 a.m., "Hannes Schmelzer" <[email protected] <mailto:[email protected]>> a scris:
Quoted 4 lines · Eddy Petri?or
Quoted 2 lines
2016-02-09 19:41 GMT+02:00 Eddy Petri?or <[email protected]
<mailto:[email protected]>>:
Quoted 11 lines
Quoted 5 lines
Hannes, can you confirm that with this patch the reported size is correct and the boot is normal again?
I initially thought the size might be incorret due to some buggy HW I have (it has some unstable DDR settings, which lead to incorrect reports on my side initially).Hi Eddy,
the board does boot with this patch.
But the reported memsize is wrong.
1GiB instead 256MiB.
Quoted 6 lines
Quoted 1 line
So we need some v2 to get rid.
Taking into account the first version was reverted, would you be willing to test a reworked patch targeting v2016.05, (when I have it ready)?
Hi Eddy,
for me - yes.
But as Wolfgang said, the code is maybe already perfect :-) Do we really have here the need to tune this?
PATCH 1/1 common/memsize.c: fix endless loop when saving
Pe 10 feb. 2016 7:58 p.m., "Hannes Schmelzer" <[email protected]> a scris:
Quoted 4 lines · Eddy Petri?or
Quoted 1 line
Taking into account the first version was reverted, would you be willing
to test a reworked patch targeting v2016.05, (when I have it ready)?
Quoted 5 lines
Hi Eddy,
for me - yes.
But as Wolfgang said, the code is maybe already perfect :-) Do we really have here the need to tune this?
I was looking at that code due to my hw's DDR issues and spent some time trying to understand why does it have duplicated code, and I assume I am not the only one nor will be the last one.
But to be honest, I think the better approach would be to do the change in 2 steps, first the upper block, then the bottom one.
Eddy