Skip to contentSkip to conversation

Makefile: doesn't need check stack size when dtb is not built

PATCH New AKASHI Takahiro 2 participants 6 messages Mar 4, 2020 – Mar 5, 2020 Kbuild

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Makefile: doesn't need check stack size when dtb is not built

The commit 5fed97af20da ("Makefile: ensure DTB doesn't overflow into initial stack") adds an extra check for stack size in BSS if CONFIG_SYS_INIT_SP_BSS_OFFSET is enabled.
This check, however, doesn't make sense under the configuration where control dtb won't be built in and it should be void in such cases.

Fixes 5fed97af20da ("Makefile: ensure DTB doesn't overflow into

initial stack")

---
 Makefile | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
Makefile +8 -1
index 0af89e0a7881..78efd7e9e250 100644
@@ -1208,7 +1208,10 @@ binary_size_check: u-boot-nodtb.bin FORCE
 		fi \
 	fi
 
-ifdef CONFIG_INIT_SP_RELATIVE
+ifeq ($(CONFIG_INIT_SP_RELATIVE),y)
+ifneq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_EMBED)$(CONFIG_OF_HOSTFILE),)
+# u-boot.dtb will be built only if one of those options is enabled
+
 ifneq ($(CONFIG_SYS_MALLOC_F_LEN),)
 subtract_sys_malloc_f_len = space=$$(($${space} - $(CONFIG_SYS_MALLOC_F_LEN)))
 else
@@ -1233,6 +1236,10 @@ init_sp_bss_offset_check: u-boot.dtb FORCE
 		echo "(CONFIG_SYS_INIT_SP_BSS_OFFSET - CONFIG_SYS_MALLOC_F_LEN)" >&2 ; \
 		exit 1 ; \
 	fi
+else
+init_sp_bss_offset_check:
+
+endif
 endif
 
 u-boot-nodtb.bin: u-boot FORCE
Signature · 1 line
2.25.1
#1 lore raw

PATCH 1/1 Makefile: doesn't need check stack size when dtb is not built

Quoted 5 lines · AKASHI Takahiro

The commit 5fed97af20da ("Makefile: ensure DTB doesn't overflow into initial stack") adds an extra check for stack size in BSS if CONFIG_SYS_INIT_SP_BSS_OFFSET is enabled.
This check, however, doesn't make sense under the configuration where control dtb won't be built in and it should be void in such cases.

Don't you want to edit the following hunk from the original patch instead or as well?

+ifneq ($(CONFIG_SYS_INIT_SP_BSS_OFFSET),)
+ALL-y += init_sp_bss_offset_check
+endif

That's why there's no rule for init_sp_bss_offset_check in the else case.

#2 lore raw

PATCH 1/1 Makefile: doesn't need check stack size when dtb is not built

Quoted 14 lines · Stephen Warren
Quoted 5 lines · AKASHI Takahiro

The commit 5fed97af20da ("Makefile: ensure DTB doesn't overflow into initial stack") adds an extra check for stack size in BSS if CONFIG_SYS_INIT_SP_BSS_OFFSET is enabled.
This check, however, doesn't make sense under the configuration where control dtb won't be built in and it should be void in such cases.

Don't you want to edit the following hunk from the original patch instead or as well?

(quoted diff) +3 -0
+ifneq ($(CONFIG_SYS_INIT_SP_BSS_OFFSET),)
+ALL-y += init_sp_bss_offset_check
+endif
Quoted 1 line

That's why there's no rule for init_sp_bss_offset_check in the else case.

I intentionally left it as it is because someone may in the future want to add other *sanity checks* whether dtb is used or not. Rather, my concern is:
Is "ifneq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_EMBED)$(CONFIG_OF_HOSTFILE),)" sufficient and appropriate to guard the check?

Thanks,
-Takahiro Akashi

#3 lore raw

PATCH 1/1 Makefile: doesn't need check stack size when dtb is not built

Quoted 11 lines · AKASHI Takahiro
Quoted 10 lines · Stephen Warren
Quoted 5 lines · AKASHI Takahiro

The commit 5fed97af20da ("Makefile: ensure DTB doesn't overflow into initial stack") adds an extra check for stack size in BSS if CONFIG_SYS_INIT_SP_BSS_OFFSET is enabled.
This check, however, doesn't make sense under the configuration where control dtb won't be built in and it should be void in such cases.

Don't you want to edit the following hunk from the original patch instead or as well?

Quoted 8 lines
Quoted 5 lines
(quoted diff) +3 -0
+ifneq ($(CONFIG_SYS_INIT_SP_BSS_OFFSET),)
+ALL-y += init_sp_bss_offset_check
+endif

That's why there's no rule for init_sp_bss_offset_check in the else case.

I intentionally left it as it is because someone may in the future want to add other *sanity checks* whether dtb is used or not.

I'd probably expect the following in that case:

ifneq ($(CONFIG_SYS_INIT_SP_BSS_OFFSET),)
ifneq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_EMBED)$(CONFIG_OF_HOSTFILE),)

     ALL-y += init_sp_bss_offset_check

endif
ALL-y += some_other_check
endif

Quoted 3 lines

Rather, my concern is:
Is "ifneq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_EMBED)$(CONFIG_OF_HOSTFILE),)" sufficient and appropriate to guard the check?

I think OF_SEPARATE is the only case this check makes sense. OF_EMBED sounds like the build process puts the DTB into .data or similar, so the issue doesn't apply. Since OF_HOSTFILE is a sandbox thing where the DT is read from a file, the check definitely doesn't apply.

#4 lore raw

PATCH 1/1 Makefile: doesn't need check stack size when dtb is not built

Quoted 30 lines · Stephen Warren
Quoted 11 lines · AKASHI Takahiro
Quoted 10 lines · Stephen Warren
Quoted 5 lines · AKASHI Takahiro

The commit 5fed97af20da ("Makefile: ensure DTB doesn't overflow into initial stack") adds an extra check for stack size in BSS if CONFIG_SYS_INIT_SP_BSS_OFFSET is enabled.
This check, however, doesn't make sense under the configuration where control dtb won't be built in and it should be void in such cases.

Don't you want to edit the following hunk from the original patch instead or as well?

Quoted 8 lines
Quoted 5 lines
(quoted diff) +3 -0
+ifneq ($(CONFIG_SYS_INIT_SP_BSS_OFFSET),)
+ALL-y += init_sp_bss_offset_check
+endif

That's why there's no rule for init_sp_bss_offset_check in the else case.

I intentionally left it as it is because someone may in the future want to add other *sanity checks* whether dtb is used or not.

I'd probably expect the following in that case:

ifneq ($(CONFIG_SYS_INIT_SP_BSS_OFFSET),)
ifneq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_EMBED)$(CONFIG_OF_HOSTFILE),)

    ALL-y += init_sp_bss_offset_check

endif
ALL-y += some_other_check
endif

Quoted 8 lines
Quoted 3 lines

Rather, my concern is:
Is "ifneq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_EMBED)$(CONFIG_OF_HOSTFILE),)" sufficient and appropriate to guard the check?

I think OF_SEPARATE is the only case this check makes sense. OF_EMBED sounds like the build process puts the DTB into .data or similar, so the issue doesn't apply. Since OF_HOSTFILE is a sandbox thing where the DT is read from a file, the check definitely doesn't apply.

Okay, sounds reasonable.
(and the target should be dts/dt.dtb rather than u-boot.dtb for clarity?)

Thanks,
-Takahiro Akashi

#5 lore raw

PATCH 1/1 Makefile: doesn't need check stack size when dtb is not built

Quoted 35 lines · AKASHI Takahiro
Quoted 34 lines · Stephen Warren
Quoted 11 lines · AKASHI Takahiro
Quoted 10 lines · Stephen Warren
Quoted 5 lines · AKASHI Takahiro

The commit 5fed97af20da ("Makefile: ensure DTB doesn't overflow into initial stack") adds an extra check for stack size in BSS if CONFIG_SYS_INIT_SP_BSS_OFFSET is enabled.
This check, however, doesn't make sense under the configuration where control dtb won't be built in and it should be void in such cases.

Don't you want to edit the following hunk from the original patch instead or as well?

Quoted 8 lines
Quoted 5 lines
(quoted diff) +3 -0
+ifneq ($(CONFIG_SYS_INIT_SP_BSS_OFFSET),)
+ALL-y += init_sp_bss_offset_check
+endif

That's why there's no rule for init_sp_bss_offset_check in the else case.

I intentionally left it as it is because someone may in the future want to add other *sanity checks* whether dtb is used or not.

I'd probably expect the following in that case:

ifneq ($(CONFIG_SYS_INIT_SP_BSS_OFFSET),)
ifneq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_EMBED)$(CONFIG_OF_HOSTFILE),)

     ALL-y += init_sp_bss_offset_check

endif
ALL-y += some_other_check
endif

Quoted 3 lines

Rather, my concern is:
Is "ifneq ($(CONFIG_OF_SEPARATE)$(CONFIG_OF_EMBED)$(CONFIG_OF_HOSTFILE),)" sufficient and appropriate to guard the check?

Quoted 7 lines
Quoted 4 lines

I think OF_SEPARATE is the only case this check makes sense. OF_EMBED sounds like the build process puts the DTB into .data or similar, so the issue doesn't apply. Since OF_HOSTFILE is a sandbox thing where the DT is read from a file, the check definitely doesn't apply.

Okay, sounds reasonable.
(and the target should be dts/dt.dtb rather than u-boot.dtb for clarity?)

By "target" I assume you mean the DTB filename in the following?

	@dtb_size=$(shell wc -c u-boot.dtb | awk '{print $$1}') ; \

If so, it doesn't make any difference; u-boot.dtb is just a copy of dts/dt.dtb:

u-boot.dtb: dts/dt.dtb

	$(call cmd,copy)