powerpc/mpc85xx: Increase image size
Scott, et al.
I'd like to start the discussion to increase u-boot image size for some mpc85xx targets. As we all know the reset vector is at the very end and linking process start from the top. This gives us no good choice but to use fixed image size. While we have more and more features, the size increases inevitably. It's time to adjust the arbitrary size. We are now using 512KB. Shall we go with 768KB, or even 1MB?
For the first step, I think we don't have to increase size for all targets. We can adjust those with CONFIG_SYS_TEXT_BASE=0xeff80000. Those are the most recent used. There are other targets which don't use NOR flash boot method. They should be considered as well.
It is per board configuration. But it may be better if we keep them consistent.
Please comment.
Regards,
York
Quoted 8 lines · York Sun
Scott, et al.
I'd like to start the discussion to increase u-boot image size for some mpc85xx targets. As we all know the reset vector is at the very end and linking process start from the top. This gives us no good choice but to use fixed image size. While we have more and more features, the size increases inevitably. It's time to adjust the arbitrary size. We are now using 512KB. Shall we go with 768KB, or even 1MB?
768K would affect fewer existing partition maps (many of which leave 1M for U-Boot and environment combined), but 1M might be better for new boards. And of course it would be nice if someone could spare some time to try to slim things down (finer-grained compile-time config, speed/size tradeoffs, etc).
Quoted 7 lines
For the first step, I think we don't have to increase size for all targets. We can adjust those with
CONFIG_SYS_TEXT_BASE=0xeff80000. Those are the most recent used. There are other targets which don't use NOR flash boot method. They should be considered as well.It is per board configuration. But it may be better if we keep them consistent.
I don't think it's worth trying to keep them consistent. Leave alone old boards that are not pushing the limit, and where testing and user education would be a hassle, and let newer boards where more features are wanted not be constrained by the past.
-Scott
Quoted 21 lines · Scott Wood
Quoted 8 lines · York Sun
Scott, et al.
I'd like to start the discussion to increase u-boot image size for some mpc85xx targets. As we all know the reset vector is at the very end and linking process start from the top. This gives us no good choice but to use fixed image size. While we have more and more features, the size increases inevitably. It's time to adjust the arbitrary size. We are now using 512KB. Shall we go with 768KB, or even 1MB?
768K would affect fewer existing partition maps (many of which leave 1M for U-Boot and environment combined), but 1M might be better for new boards. And of course it would be nice if someone could spare some time to try to slim things down (finer-grained compile-time config, speed/size tradeoffs, etc).
Quoted 5 lines
For the first step, I think we don't have to increase size for all targets. We can adjust those with
CONFIG_SYS_TEXT_BASE=0xeff80000. Those are the most recent used. There are other targets which don't use NOR flash boot method. They should be considered as well.
Quoted 8 lines
Quoted 2 lines
It is per board configuration. But it may be better if we keep them consistent.
I don't think it's worth trying to keep them consistent. Leave alone old boards that are not pushing the limit, and where testing and user education would be a hassle, and let newer boards where more features are wanted not be constrained by the past.
I did a quick search for eff80000. Only these boards have it
include/configs/B4860QDS.hinclude/configs/C29XPCIE.hinclude/configs/corenet_ds.hinclude/configs/HWW1U1A.hinclude/configs/MPC8536DS.hinclude/configs/MPC8572DS.hinclude/configs/P1010RDB.hinclude/configs/P1022DS.hinclude/configs/P1023RDB.hinclude/configs/P1023RDS.hinclude/configs/P1_P2_RDB.hinclude/configs/p1_p2_rdb_pc.hinclude/configs/p1_twr.hinclude/configs/P2020DS.hinclude/configs/P2041RDB.hinclude/configs/T1040QDS.hinclude/configs/t4qds.h
York
Quoted 42 lines · York Sun
Quoted 21 lines · Scott Wood
Quoted 8 lines · York Sun
Scott, et al.
I'd like to start the discussion to increase u-boot image size for some mpc85xx targets. As we all know the reset vector is at the very end and linking process start from the top. This gives us no good choice but to use fixed image size. While we have more and more features, the size increases inevitably. It's time to adjust the arbitrary size. We are now using 512KB. Shall we go with 768KB, or even 1MB?
768K would affect fewer existing partition maps (many of which leave 1M for U-Boot and environment combined), but 1M might be better for new boards. And of course it would be nice if someone could spare some time to try to slim things down (finer-grained compile-time config, speed/size tradeoffs, etc).
Quoted 5 lines
For the first step, I think we don't have to increase size for all targets. We can adjust those with
CONFIG_SYS_TEXT_BASE=0xeff80000. Those are the most recent used. There are other targets which don't use NOR flash boot method. They should be considered as well.Quoted 8 lines
Quoted 2 lines
It is per board configuration. But it may be better if we keep them consistent.
I don't think it's worth trying to keep them consistent. Leave alone old boards that are not pushing the limit, and where testing and user education would be a hassle, and let newer boards where more features are wanted not be constrained by the past.
I did a quick search for
eff80000. Only these boards have it
include/configs/B4860QDS.hinclude/configs/C29XPCIE.hinclude/configs/corenet_ds.hinclude/configs/HWW1U1A.hinclude/configs/MPC8536DS.hinclude/configs/MPC8572DS.hinclude/configs/P1010RDB.hinclude/configs/P1022DS.hinclude/configs/P1023RDB.h
Quoted 9 lines
include/configs/P1023RDS.hinclude/configs/P1_P2_RDB.hinclude/configs/p1_p2_rdb_pc.hinclude/configs/p1_twr.hinclude/configs/P2020DS.hinclude/configs/P2041RDB.hinclude/configs/T1040QDS.hinclude/configs/t4qds.h
Scott,
Are SPL and TPL boot methods immune from the size issue here?
York
Quoted 53 lines · York Sun
Quoted 42 lines · York Sun
Quoted 21 lines · Scott Wood
Quoted 8 lines · York Sun
Scott, et al.
I'd like to start the discussion to increase u-boot image size for some mpc85xx targets. As we all know the reset vector is at the very end and linking process start from the top. This gives us no good choice but to use fixed image size. While we have more and more features, the size increases inevitably. It's time to adjust the arbitrary size. We are now using 512KB. Shall we go with 768KB, or even 1MB?
768K would affect fewer existing partition maps (many of which leave 1M for U-Boot and environment combined), but 1M might be better for new boards. And of course it would be nice if someone could spare some time to try to slim things down (finer-grained compile-time config, speed/size tradeoffs, etc).
Quoted 5 lines
For the first step, I think we don't have to increase size for all targets. We can adjust those with
CONFIG_SYS_TEXT_BASE=0xeff80000. Those are the most recent used. There are other targets which don't use NOR flash boot method. They should be considered as well.Quoted 8 lines
Quoted 2 lines
It is per board configuration. But it may be better if we keep them consistent.
I don't think it's worth trying to keep them consistent. Leave alone old boards that are not pushing the limit, and where testing and user education would be a hassle, and let newer boards where more features are wanted not be constrained by the past.
I did a quick search for
eff80000. Only these boards have it
include/configs/B4860QDS.hinclude/configs/C29XPCIE.hinclude/configs/corenet_ds.hinclude/configs/HWW1U1A.hinclude/configs/MPC8536DS.hinclude/configs/MPC8572DS.hinclude/configs/P1010RDB.hinclude/configs/P1022DS.hinclude/configs/P1023RDB.hQuoted 9 lines
include/configs/P1023RDS.hinclude/configs/P1_P2_RDB.hinclude/configs/p1_p2_rdb_pc.hinclude/configs/p1_twr.hinclude/configs/P2020DS.hinclude/configs/P2041RDB.hinclude/configs/T1040QDS.hinclude/configs/t4qds.h
Quoted 3 lines
Scott,
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
-Scott
Quoted 54 lines · Scott Wood
Quoted 53 lines · York Sun
Quoted 51 lines · York Sun
Quoted 21 lines · Scott Wood
Quoted 8 lines · York Sun
Scott, et al.
I'd like to start the discussion to increase u-boot image size for some mpc85xx targets. As we all know the reset vector is at the very end and linking process start from the top. This gives us no good choice but to use fixed image size. While we have more and more features, the size increases inevitably. It's time to adjust the arbitrary size. We are now using 512KB. Shall we go with 768KB, or even 1MB?
768K would affect fewer existing partition maps (many of which leave 1M for U-Boot and environment combined), but 1M might be better for new boards. And of course it would be nice if someone could spare some time to try to slim things down (finer-grained compile-time config, speed/size tradeoffs, etc).
Quoted 5 lines
For the first step, I think we don't have to increase size for all targets. We can adjust those with
CONFIG_SYS_TEXT_BASE=0xeff80000. Those are the most recent used. There are other targets which don't use NOR flash boot method. They should be considered as well.Quoted 8 lines
Quoted 2 lines
It is per board configuration. But it may be better if we keep them consistent.
I don't think it's worth trying to keep them consistent. Leave alone old boards that are not pushing the limit, and where testing and user education would be a hassle, and let newer boards where more features are wanted not be constrained by the past.
I did a quick search for
eff80000. Only these boards have it
include/configs/B4860QDS.hinclude/configs/C29XPCIE.hinclude/configs/corenet_ds.hinclude/configs/HWW1U1A.hinclude/configs/MPC8536DS.hinclude/configs/MPC8572DS.hinclude/configs/P1010RDB.hinclude/configs/P1022DS.hinclude/configs/P1023RDB.hinclude/configs/P1023RDS.hinclude/configs/P1_P2_RDB.hinclude/configs/p1_p2_rdb_pc.hinclude/configs/p1_twr.hinclude/configs/P2020DS.hinclude/configs/P2041RDB.hinclude/configs/T1040QDS.hinclude/configs/t4qds.h
Quoted 6 lines
Quoted 3 lines
Scott,
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
PBL boot will be broken if the image size is bigger than 512KB, right?
York
Quoted 55 lines · York Sun
Quoted 54 lines · Scott Wood
Quoted 53 lines · York Sun
Quoted 51 lines · York Sun
Quoted 21 lines · Scott Wood
Quoted 8 lines · York Sun
Scott, et al.
I'd like to start the discussion to increase u-boot image size for some mpc85xx targets. As we all know the reset vector is at the very end and linking process start from the top. This gives us no good choice but to use fixed image size. While we have more and more features, the size increases inevitably. It's time to adjust the arbitrary size. We are now using 512KB. Shall we go with 768KB, or even 1MB?
768K would affect fewer existing partition maps (many of which leave 1M for U-Boot and environment combined), but 1M might be better for new boards. And of course it would be nice if someone could spare some time to try to slim things down (finer-grained compile-time config, speed/size tradeoffs, etc).
Quoted 5 lines
For the first step, I think we don't have to increase size for all targets. We can adjust those with
CONFIG_SYS_TEXT_BASE=0xeff80000. Those are the most recent used. There are other targets which don't use NOR flash boot method. They should be considered as well.Quoted 8 lines
Quoted 2 lines
It is per board configuration. But it may be better if we keep them consistent.
I don't think it's worth trying to keep them consistent. Leave alone old boards that are not pushing the limit, and where testing and user education would be a hassle, and let newer boards where more features are wanted not be constrained by the past.
I did a quick search for
eff80000. Only these boards have it
include/configs/B4860QDS.hinclude/configs/C29XPCIE.hinclude/configs/corenet_ds.hinclude/configs/HWW1U1A.hinclude/configs/MPC8536DS.hinclude/configs/MPC8572DS.hinclude/configs/P1010RDB.hinclude/configs/P1022DS.hinclude/configs/P1023RDB.hinclude/configs/P1023RDS.hinclude/configs/P1_P2_RDB.hinclude/configs/p1_p2_rdb_pc.hinclude/configs/p1_twr.hinclude/configs/P2020DS.hinclude/configs/P2041RDB.hinclude/configs/T1040QDS.hinclude/configs/t4qds.h
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
Scott,
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
PBL boot will be broken if the image size is bigger than 512KB, right?
It has to be even smaller than that, to make room for early data.
-Scott
Quoted 58 lines · Scott Wood
Quoted 57 lines · York Sun
Quoted 56 lines · Scott Wood
Quoted 55 lines · York Sun
Quoted 42 lines · York Sun
Quoted 21 lines · Scott Wood
Quoted 8 lines · York Sun
Scott, et al.
I'd like to start the discussion to increase u-boot image size for some mpc85xx targets. As we all know the reset vector is at the very end and linking process start from the top. This gives us no good choice but to use fixed image size. While we have more and more features, the size increases inevitably. It's time to adjust the arbitrary size. We are now using 512KB. Shall we go with 768KB, or even 1MB?
768K would affect fewer existing partition maps (many of which leave 1M for U-Boot and environment combined), but 1M might be better for new boards. And of course it would be nice if someone could spare some time to try to slim things down (finer-grained compile-time config, speed/size tradeoffs, etc).
Quoted 5 lines
For the first step, I think we don't have to increase size for all targets. We can adjust those with
CONFIG_SYS_TEXT_BASE=0xeff80000. Those are the most recent used. There are other targets which don't use NOR flash boot method. They should be considered as well.Quoted 8 lines
Quoted 2 lines
It is per board configuration. But it may be better if we keep them consistent.
I don't think it's worth trying to keep them consistent. Leave alone old boards that are not pushing the limit, and where testing and user education would be a hassle, and let newer boards where more features are wanted not be constrained by the past.
I did a quick search for
eff80000. Only these boards have it
include/configs/B4860QDS.hinclude/configs/C29XPCIE.hinclude/configs/corenet_ds.hinclude/configs/HWW1U1A.hinclude/configs/MPC8536DS.hinclude/configs/MPC8572DS.hinclude/configs/P1010RDB.hinclude/configs/P1022DS.hinclude/configs/P1023RDB.hQuoted 9 lines
include/configs/P1023RDS.hinclude/configs/P1_P2_RDB.hinclude/configs/p1_p2_rdb_pc.hinclude/configs/p1_twr.hinclude/configs/P2020DS.hinclude/configs/P2041RDB.hinclude/configs/T1040QDS.hinclude/configs/t4qds.hScott,
Quoted 9 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
PBL boot will be broken if the image size is bigger than 512KB, right?
It has to be even smaller than that, to make room for early data.
So if we go with 768KB, do we have to convert all PBL boot to SPL boot?
York
Quoted 9 lines · York Sun
Quoted 8 lines · Scott Wood
Quoted 7 lines · York Sun
Quoted 5 lines · Scott Wood
Quoted 1 line · York Sun
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
Quoted 6 lines
Quoted 4 lines
Quoted 1 line
PBL boot will be broken if the image size is bigger than 512KB, right?
It has to be even smaller than that, to make room for early data.
So if we go with 768KB, do we have to convert all PBL boot to SPL boot?
Only the targets that need the extra space.
-Scott
Quoted 10 lines · Scott Wood
Quoted 9 lines · York Sun
Quoted 8 lines · Scott Wood
Quoted 7 lines · York Sun
Quoted 5 lines · Scott Wood
Quoted 1 line · York Sun
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
Quoted 9 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line
PBL boot will be broken if the image size is bigger than 512KB, right?
It has to be even smaller than that, to make room for early data.
So if we go with 768KB, do we have to convert all PBL boot to SPL boot?
Only the targets that need the extra space.
We have T4, B4 and corenet_ds using PBL boot. They most likely will exceed the 512KB soon, if not yet. It maybe easier to change all of them togther, than one by one.
York
Quoted 16 lines · York Sun
Quoted 15 lines · Scott Wood
Quoted 14 lines · York Sun
Quoted 12 lines · Scott Wood
Quoted 8 lines · York Sun
Quoted 5 lines · Scott Wood
Quoted 1 line · York Sun
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
PBL boot will be broken if the image size is bigger than 512KB, right?
It has to be even smaller than that, to make room for early data.
Quoted 8 lines
Quoted 4 lines
Quoted 1 line
So if we go with 768KB, do we have to convert all PBL boot to SPL boot?
Only the targets that need the extra space.
We have T4, B4 and
corenet_dsusing PBL boot. They most likely will exceed the 512KB soon, if not yet. It maybe easier to change all of them togther, than one by one.
There's no reason to change them all at once. It doesn't make anything easier; it just means you have to do a bunch of testing all at once, and force a change in procedure for users on boards where it otherwise would not have been required.
Plus, the 512K limit is for e500v2-based chips. Newer chips have CPC for SRAM which is larger than 512K.
-Scott
Quoted 22 lines · Scott Wood
Quoted 21 lines · York Sun
Quoted 19 lines · Scott Wood
Quoted 9 lines · York Sun
Quoted 8 lines · Scott Wood
Quoted 7 lines · York Sun
Quoted 5 lines · Scott Wood
Quoted 1 line · York Sun
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
Quoted 6 lines
Quoted 4 lines
Quoted 1 line
PBL boot will be broken if the image size is bigger than 512KB, right?
It has to be even smaller than that, to make room for early data.
So if we go with 768KB, do we have to convert all PBL boot to SPL boot?
Only the targets that need the extra space.
Quoted 8 lines
Quoted 3 lines
We have T4, B4 and
corenet_dsusing PBL boot. They most likely will exceed the 512KB soon, if not yet. It maybe easier to change all of them togther, than one by one.There's no reason to change them all at once. It doesn't make anything easier; it just means you have to do a bunch of testing all at once, and force a change in procedure for users on boards where it otherwise would not have been required.
You are right here.
Quoted 4 lines
Plus, the 512K limit is for e500v2-based chips. Newer chips have CPC for SRAM which is larger than 512K.
Hmm? T4240 has 512KB CPC.
York
Quoted 32 lines · York Sun
Quoted 22 lines · Scott Wood
Quoted 21 lines · York Sun
Quoted 19 lines · Scott Wood
Quoted 9 lines · York Sun
Quoted 8 lines · Scott Wood
Quoted 7 lines · York Sun
Quoted 5 lines · Scott Wood
Quoted 1 line · York Sun
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
Quoted 6 lines
Quoted 4 lines
Quoted 1 line
PBL boot will be broken if the image size is bigger than 512KB, right?
It has to be even smaller than that, to make room for early data.
So if we go with 768KB, do we have to convert all PBL boot to SPL boot?
Only the targets that need the extra space.
Quoted 8 lines
Quoted 3 lines
We have T4, B4 and
corenet_dsusing PBL boot. They most likely will exceed the 512KB soon, if not yet. It maybe easier to change all of them togther, than one by one.There's no reason to change them all at once. It doesn't make anything easier; it just means you have to do a bunch of testing all at once, and force a change in procedure for users on boards where it otherwise would not have been required.
Quoted 8 lines
You are right here.
Quoted 4 lines
Plus, the 512K limit is for e500v2-based chips. Newer chips have CPC for SRAM which is larger than 512K.
Hmm? T4240 has 512KB CPC.
Each CPC is 512K, but there are three of them. Is it possible to use more than one for SRAM, contiguously?
Still, I find it curious that T4240 has less CPC than P4080, and less CPC than L2 cache.
-Scott
Quoted 36 lines · Scott Wood
Quoted 35 lines · York Sun
Quoted 22 lines · Scott Wood
Quoted 21 lines · York Sun
Quoted 19 lines · Scott Wood
Quoted 9 lines · York Sun
Quoted 8 lines · Scott Wood
Quoted 7 lines · York Sun
Quoted 5 lines · Scott Wood
Quoted 1 line · York Sun
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition tables.
Quoted 6 lines
Quoted 4 lines
Quoted 1 line
PBL boot will be broken if the image size is bigger than 512KB, right?
It has to be even smaller than that, to make room for early data.
So if we go with 768KB, do we have to convert all PBL boot to SPL boot?
Only the targets that need the extra space.
Quoted 8 lines
Quoted 3 lines
We have T4, B4 and
corenet_dsusing PBL boot. They most likely will exceed the 512KB soon, if not yet. It maybe easier to change all of them togther, than one by one.There's no reason to change them all at once. It doesn't make anything easier; it just means you have to do a bunch of testing all at once, and force a change in procedure for users on boards where it otherwise would not have been required.
You are right here.
Quoted 1 line
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Plus, the 512K limit is for e500v2-based chips. Newer chips have CPC for SRAM which is larger than 512K.
Hmm? T4240 has 512KB CPC.
Each CPC is 512K, but there are three of them. Is it possible to use more than one for SRAM, contiguously?
I think it is possible to use two CPC. The PBL boot method is hard-coded to take 512KB u-boot.bin and generate u-boot.pbl. It surely can be improved. My point is making a move for all concerned platforms may be easier for maintenance.
York
Quoted 56 lines
-----Original Message-----
From: u-boot-bounces at lists.denx.de [mailto:u-boot-bounces at lists.denx.de] On Behalf Of York Sun
Sent: Thursday, October 17, 2013 4:14 AM
To: Wood Scott-B07421
Cc: Tom Rini; u-boot at lists.denx.de
Subject: Re: [U-Boot] powerpc/mpc85xx: Increase image sizeQuoted 4 lines · Scott Wood
Quoted 3 lines · York Sun
Quoted 2 lines · Scott Wood
Quoted 1 line · York Sun
On 10/16/2013 01:33 PM, Scott Wood wrote:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines · York Sun
Quoted 7 lines · Scott Wood
Quoted 6 lines · York Sun
Quoted 5 lines · Scott Wood
Quoted 4 lines · York Sun
Quoted 1 line
Are SPL and TPL boot methods immune from the size issue here?
Sort of. We still need to fit inside existing partition
tables.
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line
PBL boot will be broken if the image size is bigger than 512KB,
right?
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 3 lines
It has to be even smaller than that, to make room for early data.
So if we go with 768KB, do we have to convert all PBL boot to SPL
boot?
Quoted 22 lines
Quoted 21 lines
Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Only the targets that need the extra space.
Quoted 8 lines
Quoted 3 lines
We have T4, B4 and
corenet_dsusing PBL boot. They most likely will exceed the 512KB soon, if not yet. It maybe easier to change all of them togther, than one by one.There's no reason to change them all at once. It doesn't make anything easier; it just means you have to do a bunch of testing all at once, and force a change in procedure for users on boards where it otherwise would not have been required.
You are right here.
Quoted 4 lines
Plus, the 512K limit is for e500v2-based chips. Newer chips have CPC for SRAM which is larger than 512K.
Hmm? T4240 has 512KB CPC.
Quoted 8 lines
Quoted 2 lines
Each CPC is 512K, but there are three of them. Is it possible to use more than one for SRAM, contiguously?
I think it is possible to use two CPC. The PBL boot method is hard-coded to take 512KB u-boot.bin and generate u-boot.pbl. It surely can be improved. My point is making a move for all concerned platforms may be easier for maintenance.
We can go ahead with 2 stage boot loaders for SoC having < 1MB CPC.
We have already sent patch for T1040 which has 256K CPC. This same patch can be used for T4 and B4.
FYI, I have also sent one RFC for B4 with 2stage boot loader. It can be taken as full patch.
Regards,
Prabhakar