ARC buildroot fixes/updates
cover ARC buildroot fixes/updates
Hi,
I have some fixes/updates from ongoing glibc work. Please review/apply.
Thx,
-Vineet
Vineet Gupta (3):
arch/config.in.arc: Introduce the ARC optimized hs38 config arch/config.in.arc: Introduce ARC ISA toggle to ease downstream
toggles
package/ffmpeg: Enable ARC glibc builds
arch/Config.in.arc | 44 ++++++++++++++++++++++++++++++++-------- package/ffmpeg/Config.in | 2 -- 2 files changed, 36 insertions(+), 10 deletions(-)
Signature · 1 line
2.20.1
PATCH 1/3 arch/config.in.arc: Introduce the ARC optimized hs38 config
This corresponds to -mcu=hs38 with mpy-option=9 (64-bit multiplier)
--- arch/Config.in.arc | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-)
arch/Config.in.arc +14 -7
@@ -11,13 +11,19 @@ config BR2_arc750d config BR2_arc770d bool "ARC 770D" -config BR2_archs38 +config BR2_archs bool "ARC HS38" help Generic ARC HS capable of running Linux, i.e. with MMU, - caches and multiplier. Also it corresponds to the default + caches and 32-bit multiplier. Also it corresponds to the default configuration in older GNU toolchain versions. +config BR2_archs38 + bool "ARC HS38 with 64-bit mpy" + help + Fully featured ARC HS capable of running Linux, i.e. with MMU, + caches and 64-bit multiplier. + If you're not sure which version of ARC HS core you build for keep this one.
@@ -43,7 +49,7 @@ endchoice # Choice of atomic instructions presence config BR2_ARC_ATOMIC_EXT bool "Atomic extension (LLOCK/SCOND instructions)" - default y if BR2_arc770d || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 + default y if BR2_arc770d || BR2_archs || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 config BR2_ARCH default "arc" if BR2_arcle
@@ -60,13 +66,14 @@ config BR2_ENDIAN config BR2_GCC_TARGET_CPU default "arc700" if BR2_arc750d default "arc700" if BR2_arc770d - default "archs" if BR2_archs38 + default "archs" if BR2_archs + default "hs38" if BR2_archs38 default "hs38_linux" if BR2_archs38_full default "hs4x_rel31" if BR2_archs4x_rel31 config BR2_READELF_ARCH_NAME default "ARCompact" if BR2_arc750d || BR2_arc770d - default "ARCv2" if BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 + default "ARCv2" if BR2_archs || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 choice prompt "MMU Page Size"
@@ -86,7 +93,7 @@ choice config BR2_ARC_PAGE_SIZE_4K bool "4KB" - depends on BR2_arc770d || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 + depends on BR2_arc770d || BR2_archs || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 config BR2_ARC_PAGE_SIZE_8K bool "8KB"
@@ -96,7 +103,7 @@ config BR2_ARC_PAGE_SIZE_8K config BR2_ARC_PAGE_SIZE_16K bool "16KB" - depends on BR2_arc770d || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 + depends on BR2_arc770d || BR2_archs || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 endchoice
Signature · 1 line
2.20.1
PATCH 1/3 arch/config.in.arc: Introduce the ARC optimized hs38 config
Hello,
+Arnout for legacy handling.
Quoted 15 lines · On Fri, 8 Nov 2019 09:41:10 -0800 Vineet Gupta <[email protected]>…
This corresponds to -mcu=hs38 with mpy-option=9 (64-bit multiplier)
Signed-off-by Vineet Gupta <[email protected]>--- arch/Config.in.arc | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-)arch/Config.in.arc +0
-0@@ -11,13 +11,19 @@ config BR2_arc750d config BR2_arc770d bool "ARC 770D"
arch/Config.in.arc config BR2_arc750d · line 21
Quoted 10 lines
arch/Config.in.arc +3
-2-config BR2_archs38 +config BR2_archs bool "ARC HS38" help Generic ARC HS capable of running Linux, i.e. with MMU, - caches and multiplier. Also it corresponds to the default + caches and 32-bit multiplier. Also it corresponds to the default configuration in older GNU toolchain versions. +config BR2_archs38
This re-use of an existing name is a bit annoying. Indeed, all existing users of Buildroot that have a configuration with BR2_archs38 will now be building for a ARC system with a 64-bit multiplier, while they were previously building for a 32-bit multiplier.
I see that what you have done is to try to be consistent between the BR2_ options and the gcc options. I'm hesitating between keeping the consistency but making the migration a bit annoying for users, or breaking the consistency to make the migration smooth for users.
Since I think the number of affected users will probably be quite small/limited, I think I would be fine with merging your patch as-is, but I'd like to hear from others.
Thanks,
Thomas
Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
PATCH 1/3 arch/config.in.arc: Introduce the ARC optimized hs38 config
Vineet, Thomas, All,
On 2019-11-09 14:46 +0100, Thomas Petazzoni spake thusly:
Quoted 36 lines
+Arnout for legacy handling.
Quoted 15 lines · On Fri, 8 Nov 2019 09:41:10 -0800 Vineet Gupta <[email protected]>…
This corresponds to -mcu=hs38 with mpy-option=9 (64-bit multiplier)
Signed-off-by Vineet Gupta <[email protected]>--- arch/Config.in.arc | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-)arch/Config.in.arc +0
-0@@ -11,13 +11,19 @@ config BR2_arc750d config BR2_arc770d bool "ARC 770D"arch/Config.in.arc config BR2_arc750d · line 21
Quoted 10 lines
arch/Config.in.arc +3
-2-config BR2_archs38 +config BR2_archs bool "ARC HS38" help Generic ARC HS capable of running Linux, i.e. with MMU, - caches and multiplier. Also it corresponds to the default + caches and 32-bit multiplier. Also it corresponds to the default configuration in older GNU toolchain versions. +config BR2_archs38This re-use of an existing name is a bit annoying. Indeed, all existing users of Buildroot that have a configuration with
BR2_archs38will now be building for a ARC system with a 64-bit multiplier, while they were previously building for a 32-bit multiplier.
Quoted 8 lines
I see that what you have done is to try to be consistent between the
BR2_options and the gcc options. I'm hesitating between keeping the consistency but making the migration a bit annoying for users, or breaking the consistency to make the migration smooth for users.Since I think the number of affected users will probably be quite small/limited, I think I would be fine with merging your patch as-is, but I'd like to hear from others.
I would prefer if we kept the existing name as-is, and introduce the new variant under a different name, e.g.:
config BR2_archs38
bool "ARC HS38"
config BR2_archs68_64mpy
bool "ARC HS38 w/ 64-bit mpy"Because, well, they both are HS38 cores, so we are not wrong in naming both options with hs38.
It would have been good that the config names match the gcc option, of course, but that is a minor detail I believe...
Regards,
Yann E. MORIN.
Quoted 11 lines
Thanks,
Thomas
Signature · 7 lines
Thomas Petazzoni, CTO, Bootlin Embedded Linux and Kernel engineering https://bootlin.com _______________________________________________ buildroot mailing list buildroot at busybox.net http://lists.busybox.net/mailman/listinfo/buildroot
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH 1/3 arch/config.in.arc: Introduce the ARC optimized hs38 config
In reply to Thomas PetazzoniHi Thomas,
Quoted 37 lines · Thomas Petazzoni
Hello,
+Arnout for legacy handling.
Quoted 15 lines · On Fri, 8 Nov 2019 09:41:10 -0800 Vineet Gupta <[email protected]>…
This corresponds to -mcu=hs38 with mpy-option=9 (64-bit multiplier)
Signed-off-by Vineet Gupta <[email protected]>--- arch/Config.in.arc | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-)arch/Config.in.arc +0
-0@@ -11,13 +11,19 @@ config BR2_arc750d config BR2_arc770d bool "ARC 770D"arch/Config.in.arc config BR2_arc750d · line 21
Quoted 10 lines
arch/Config.in.arc +3
-2-config BR2_archs38 +config BR2_archs bool "ARC HS38" help Generic ARC HS capable of running Linux, i.e. with MMU, - caches and multiplier. Also it corresponds to the default + caches and 32-bit multiplier. Also it corresponds to the default configuration in older GNU toolchain versions. +config BR2_archs38This re-use of an existing name is a bit annoying. Indeed, all existing users of Buildroot that have a configuration with
BR2_archs38will now be building for a ARC system with a 64-bit multiplier, while they were previously building for a 32-bit multiplier.
Quoted 8 lines
I see that what you have done is to try to be consistent between the
BR2_options and the gcc options. I'm hesitating between keeping the consistency but making the migration a bit annoying for users, or breaking the consistency to make the migration smooth for users.Since I think the number of affected users will probably be quite small/limited, I think I would be fine with merging your patch as-is, but I'd like to hear from others.
I agree that this might cause potential breakage, but this is not totally uncharted territory for software build config. We've been building Linux kernel with this option as default since mid 2018.
2018-09-07 00a99339f0a3 ARCv2: build: use mcpu=hs38 iso generic mcpu=archs?
Granted that kernel build is just one part of puzzle and any latent codegen issues are more likely to surface with default applied to full software stack, I would still vote for switching default to -mcpu=hs38
Thx,
-Vineet
PATCH 1/3 arch/config.in.arc: Introduce the ARC optimized hs38 config
Vineet, All,
On 2019-11-11 18:47 +0000, Vineet Gupta spake thusly:
Quoted 16 lines · Thomas Petazzoni
Quoted 15 lines
Quoted 13 lines · On Fri, 8 Nov 2019 09:41:10 -0800 Vineet Gupta <[email protected]>…
Signed-off-by Vineet Gupta <[email protected]>--- arch/Config.in.arc | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-)arch/Config.in.arc +0
-0@@ -11,13 +11,19 @@ config BR2_arc750d config BR2_arc770d bool "ARC 770D"
Quoted 2 lines
Quoted 2 lines
arch/Config.in.arc config BR2_arc750d · line 14
Quoted 2 lines
arch/Config.in.arc +1
-1-config BR2_archs38 +config BR2_archs
[Note this change, above ^^^]
Quoted 9 lines
Quoted 9 lines
arch/Config.in.arc config BR2_arc750d · line 22
Quoted 9 lines
arch/Config.in.arc +3
-1bool "ARC HS38" help Generic ARC HS capable of running Linux, i.e. with MMU, - caches and multiplier. Also it corresponds to the default + caches and 32-bit multiplier. Also it corresponds to the default configuration in older GNU toolchain versions. +config BR2_archs38 + bool "ARC HS38 with 64-bit mpy"
[and note this, above^^^]
Quoted 10 lines
Quoted 10 lines
This re-use of an existing name is a bit annoying. Indeed, all existing users of Buildroot that have a configuration with
BR2_archs38will now be building for a ARC system with a 64-bit multiplier, while they were previously building for a 32-bit multiplier.I see that what you have done is to try to be consistent between the
BR2_options and the gcc options. I'm hesitating between keeping the consistency but making the migration a bit annoying for users, or breaking the consistency to make the migration smooth for users.
Quoted 7 lines
Quoted 3 lines
Since I think the number of affected users will probably be quite small/limited, I think I would be fine with merging your patch as-is, but I'd like to hear from others.
I agree that this might cause potential breakage, but this is not totally uncharted territory for software build config. We've been building Linux kernel with this option as default since mid 2018.
I think there is some misunderstanding.
What Thomas and I argue on, is the way to change the meaning for the BR2_archs38 option.
Before this patch, BR2_archs38 meant "ARC HS38". But with your patch, it now means "ARC HS38 with 64-bit mpy".
So, a user that updates their Buildroot configurationi which previously had a "plain" HS38 setup, but with this patch, they get an "extended" HS38 with the 64-bit multiplier.
That's why I suggested in my own reply, to keep BR2_archs38 as it was before, and introduce BR2_archs38_64mpy to mean the new HS38 w/ the 64-bit multiplier.
Indeed, it does not match the gcc config options, but that would hardly be the only derogation we have in Buildroot... ;-)
Quoted 5 lines
2018-09-07
00a99339f0a3ARCv2: build: use mcpu=hs38 iso generic mcpu=archs?Granted that kernel build is just one part of puzzle and any latent codegen issues are more likely to surface with default applied to full software stack, I would still vote for switching default to -mcpu=hs38
Changing the meaning of an option, and changing the default of a choice, are two different things. I'm OK with changing the default, but I'd rather that options keep their meaning.
Regards,
Yann E. MORIN.
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH 1/3 arch/config.in.arc: Introduce the ARC optimized hs38 config
Hi Yann,
Quoted 11 lines · Yann E. MORIN
Quoted 11 lines
Quoted 11 lines
This re-use of an existing name is a bit annoying. Indeed, all existing users of Buildroot that have a configuration with
BR2_archs38will now be building for a ARC system with a 64-bit multiplier, while they were previously building for a 32-bit multiplier.I see that what you have done is to try to be consistent between the
BR2_options and the gcc options. I'm hesitating between keeping the consistency but making the migration a bit annoying for users, or breaking the consistency to make the migration smooth for users.
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
Since I think the number of affected users will probably be quite small/limited, I think I would be fine with merging your patch as-is, but I'd like to hear from others.
I agree that this might cause potential breakage, but this is not totally uncharted territory for software build config. We've been building Linux kernel with this option as default since mid 2018.
I think there is some misunderstanding.
Not really, I understand what you and Thomas are asking for - its not a super complicated patch after all ;-)
Quoted 2 lines
What Thomas and I argue on, is the way to change the meaning for the
BR2_archs38option.
Yeah I understand that part.
Quoted 7 lines
Before this patch,
BR2_archs38meant "ARC HS38". But with your patch, it now means "ARC HS38 with 64-bit mpy".So, a user that updates their Buildroot configurationi which previously had a "plain" HS38 setup, but with this patch, they get an "extended" HS38 with the 64-bit multiplier.
Quoted 6 lines
That's why I suggested in my own reply, to keep
BR2_archs38as it was before, and introduceBR2_archs38_64mpyto mean the new HS38 w/ the 64-bit multiplier.Indeed, it does not match the gcc config options, but that would hardly be the only derogation we have in Buildroot... ;-)
Right that's not the point of doing this exercise anyways.
Quoted 8 lines
Quoted 5 lines
2018-09-07
00a99339f0a3ARCv2: build: use mcpu=hs38 iso generic mcpu=archs?Granted that kernel build is just one part of puzzle and any latent codegen issues are more likely to surface with default applied to full software stack, I would still vote for switching default to -mcpu=hs38
Changing the meaning of an option, and changing the default of a choice, are two different things. I'm OK with changing the default, but I'd rather that options keep their meaning.
Ok sounds good !
Thx,
-Vineet
PATCH 2/3 arch/config.in.arc: Introduce ARC ISA toggle to ease downstream toggles
In reply to Vineet Gupta--- arch/Config.in.arc | 25 +++++++++++++++++++++++-- 1 file changed, 23 insertions(+), 2 deletions(-)
arch/Config.in.arc +23 -2
@@ -1,3 +1,18 @@ +choice + prompt "Target ISA" + default BR2_arcompact + depends on BR2_arc + help + Specific ARC ISA to use + +config BR2_arcompact + bool "ARCompact ISA" + +config BR2_arcv2 + bool "ARCv2 ISA" + +endchoice + choice prompt "Target CPU" default BR2_arc770d
@@ -7,12 +22,15 @@ choice config BR2_arc750d bool "ARC 750D" + depends on BR2_arcompact config BR2_arc770d bool "ARC 770D" + depends on BR2_arcompact config BR2_archs bool "ARC HS38" + depends on BR2_arcv2 help Generic ARC HS capable of running Linux, i.e. with MMU, caches and 32-bit multiplier. Also it corresponds to the default
@@ -20,6 +38,7 @@ config BR2_archs config BR2_archs38 bool "ARC HS38 with 64-bit mpy" + depends on BR2_arcv2 help Fully featured ARC HS capable of running Linux, i.e. with MMU, caches and 64-bit multiplier.
@@ -29,6 +48,7 @@ config BR2_archs38 config BR2_archs38_full bool "ARC HS38 with Quad MAC & FPU" + depends on BR2_arcv2 help Fully featured ARC HS with additional support for - Dual- and quad multiply and MC oprations
@@ -39,6 +59,7 @@ config BR2_archs38_full config BR2_archs4x_rel31 bool "ARC HS48 rel 31" + depends on BR2_arcv2 help Latest release of HS48 processor - Dual- and quad multiply and MC oprations
@@ -72,8 +93,8 @@ config BR2_GCC_TARGET_CPU default "hs4x_rel31" if BR2_archs4x_rel31 config BR2_READELF_ARCH_NAME - default "ARCompact" if BR2_arc750d || BR2_arc770d - default "ARCv2" if BR2_archs || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 + default "ARCompact" if BR2_arcompact + default "ARCv2" if BR2_arcv2 choice prompt "MMU Page Size"
Signature · 1 line
2.20.1
PATCH 2/3 arch/config.in.arc: Introduce ARC ISA toggle to ease downstream toggles
Hello,
Quoted 14 lines · On Fri, 8 Nov 2019 09:41:11 -0800 Vineet Gupta <[email protected]>…
Signed-off-by Vineet Gupta <[email protected]>--- arch/Config.in.arc | 25 +++++++++++++++++++++++-- 1 file changed, 23 insertions(+), 2 deletions(-)arch/Config.in.arc +4
-0@@ -1,3 +1,18 @@ +choice + prompt "Target ISA" + default BR2_arcompact + depends on BR2_arc
arch/Config.in.arc line 14
Quoted 10 lines
arch/Config.in.arc +10
-0+ help + Specific ARC ISA to use + +config BR2_arcompact + bool "ARCompact ISA" + +config BR2_arcv2 + bool "ARCv2 ISA" + +endchoice
I don't think we want a choice for that. It should simply be implied by the target CPU selection.
So instead, do it like this:
config BR2_ARC_ARCH_ISA_ARCOMPACT
bool
config BR2_ARC_ARCH_ISA_ARCV2
bool
(note: the names are just a proposal, there are probably some better names)
arch/Config.in.arc choice · line 25
Quoted 8 lines
arch/Config.in.arc +1
-0choice prompt "Target CPU" default BR2_arc770d@@ -7,12 +22,15 @@ choice config BR2_arc750d bool "ARC 750D" + depends on BR2_arcompact
Replace by:
select BR2_ARC_ARCH_ISA_ARCOMPACT
arch/Config.in.arc choice · line 29
Quoted 4 lines
arch/Config.in.arc +1
-0config BR2_arc770d bool "ARC 770D" + depends on BR2_arcompact
Ditto.
arch/Config.in.arc choice · line 33
Quoted 4 lines
arch/Config.in.arc +1
-0config BR2_archs bool "ARC HS38" + depends on BR2_arcv2
select BR2_ARC_ARCH_ISA_ARCV2
arch/Config.in.arc config BR2_archs · line 41
Quoted 8 lines
arch/Config.in.arc +1
-0help Generic ARC HS capable of running Linux, i.e. with MMU, caches and 32-bit multiplier. Also it corresponds to the default@@ -20,6 +38,7 @@ config BR2_archs config BR2_archs38 bool "ARC HS38 with 64-bit mpy" + depends on BR2_arcv2
select BR2_ARC_ARCH_ISA_ARCV2
arch/Config.in.arc config BR2_archs38 · line 51
Quoted 8 lines
arch/Config.in.arc +1
-0help Fully featured ARC HS capable of running Linux, i.e. with MMU, caches and 64-bit multiplier.@@ -29,6 +48,7 @@ config BR2_archs38 config BR2_archs38_full bool "ARC HS38 with Quad MAC & FPU" + depends on BR2_arcv2
select BR2_ARC_ARCH_ISA_ARCV2
arch/Config.in.arc config BR2_archs38_full · line 62
Quoted 8 lines
arch/Config.in.arc +1
-0help Fully featured ARC HS with additional support for - Dual- and quad multiply and MC oprations@@ -39,6 +59,7 @@ config BR2_archs38_full config BR2_archs4x_rel31 bool "ARC HS48 rel 31" + depends on BR2_arcv2
select BR2_ARC_ARCH_ISA_ARCV2
arch/Config.in.arc config BR2_GCC_TARGET_CPU · line 96
Quoted 10 lines
arch/Config.in.arc +1
-2help Latest release of HS48 processor - Dual- and quad multiply and MC oprations@@ -72,8 +93,8 @@ config BR2_GCC_TARGET_CPU default "hs4x_rel31" if BR2_archs4x_rel31 config BR2_READELF_ARCH_NAME - default "ARCompact" if BR2_arc750d || BR2_arc770d - default "ARCv2" if BR2_archs || BR2_archs38 || BR2_archs38_full || BR2_archs4x_rel31 + default "ARCompact" if BR2_arcompact
default "ARCompact" if BR2_ARC_ARCH_ISA_ARCOMPACT
arch/Config.in.arc config BR2_GCC_TARGET_CPU · line 97
Quoted 1 line
arch/Config.in.arc +1
-0+ default "ARCv2" if BR2_arcv2
default "ARCv2" if BR2_ARC_ARCH_ISA_ARCV2
Thanks,
Thomas
Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
PATCH 3/3 package/ffmpeg: Enable ARC glibc builds
In reply to Vineet GuptaARC glibc fenv.h has the needed macros for quite some time now so allow ffmpeg builds
--- package/ffmpeg/Config.in | 2 -- 1 file changed, 2 deletions(-)
package/ffmpeg/Config.in +0 -2
@@ -1,8 +1,6 @@ config BR2_PACKAGE_FFMPEG_ARCH_SUPPORTS bool default y - # fenv.h lacks FE_INVALID, FE_OVERFLOW & FE_UNDERFLOW - depends on !(BR2_archs38 && BR2_TOOLCHAIN_USES_GLIBC) # fenv.h lacks FE_INVALID, FE_OVERFLOW & FE_UNDERFLOW on nios2 depends on !BR2_nios2 # No support for ARMv7-M in the ARM assembly logic
Signature · 1 line
2.20.1
Thomas Petazzoni
AppliedPATCH 3/3 package/ffmpeg: Enable ARC glibc builds
Quoted 7 lines · On Fri, 8 Nov 2019 09:41:12 -0800 Vineet Gupta <[email protected]>…
ARC glibc
fenv.hhas the needed macros for quite some time now so allow ffmpeg buildsSigned-off-by Vineet Gupta <[email protected]>--- package/ffmpeg/Config.in | 2 -- 1 file changed, 2 deletions(-)
Applied to next, thanks.
Thomas
Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin Embedded Linux and Kernel engineering https://bootlin.com