[U-Boot] [PATCH] arm64 :show_regs: show the address before relocation
After relocation, when error happends, it is hard to track ELR and LR with asm file objdumped from elf file.
So subtract the gd->reloc_off the reflect the compliation address.
Signed-off-by: Peng Fan peng.fan@nxp.com --- arch/arm/lib/interrupts_64.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/arch/arm/lib/interrupts_64.c b/arch/arm/lib/interrupts_64.c index 7c9cfce69f..cbcfeec2b0 100644 --- a/arch/arm/lib/interrupts_64.c +++ b/arch/arm/lib/interrupts_64.c @@ -9,6 +9,7 @@ #include <linux/compiler.h> #include <efi_loader.h>
+DECLARE_GLOBAL_DATA_PTR;
int interrupt_init(void) { @@ -29,8 +30,13 @@ void show_regs(struct pt_regs *regs) { int i;
- printf("ELR: %lx\n", regs->elr); - printf("LR: %lx\n", regs->regs[30]); + if (gd->flags & GD_FLG_RELOC) { + printf("ELR: %lx\n", regs->elr - gd->reloc_off); + printf("LR: %lx\n", regs->regs[30] - gd->reloc_off); + } else { + printf("ELR: %lx\n", regs->elr); + printf("LR: %lx\n", regs->regs[30]); + } for (i = 0; i < 29; i += 2) printf("x%-2d: %016lx x%-2d: %016lx\n", i, regs->regs[i], i+1, regs->regs[i+1]);
On 27 November 2017 at 19:08, Peng Fan peng.fan@nxp.com wrote:
After relocation, when error happends, it is hard to track ELR and LR with asm file objdumped from elf file.
So subtract the gd->reloc_off the reflect the compliation address.
Signed-off-by: Peng Fan peng.fan@nxp.com
arch/arm/lib/interrupts_64.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
Reviewed-by: Simon Glass sjg@chromium.org
On Tue, Nov 28, 2017 at 10:08:08AM +0800, Peng Fan wrote:
After relocation, when error happends, it is hard to track ELR and LR with asm file objdumped from elf file.
So subtract the gd->reloc_off the reflect the compliation address.
Signed-off-by: Peng Fan peng.fan@nxp.com Reviewed-by: Simon Glass sjg@chromium.org
Applied to u-boot/master, thanks!
On Tue, Nov 28, 2017 at 10:08:08AM +0800, Peng Fan wrote:
After relocation, when error happends, it is hard to track ELR and LR with asm file objdumped from elf file.
So subtract the gd->reloc_off the reflect the compliation address.
Signed-off-by: Peng Fan peng.fan@nxp.com
arch/arm/lib/interrupts_64.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/arch/arm/lib/interrupts_64.c b/arch/arm/lib/interrupts_64.c index 7c9cfce69f..cbcfeec2b0 100644 --- a/arch/arm/lib/interrupts_64.c +++ b/arch/arm/lib/interrupts_64.c @@ -9,6 +9,7 @@ #include <linux/compiler.h> #include <efi_loader.h>
+DECLARE_GLOBAL_DATA_PTR;
int interrupt_init(void) { @@ -29,8 +30,13 @@ void show_regs(struct pt_regs *regs) { int i;
- printf("ELR: %lx\n", regs->elr);
- printf("LR: %lx\n", regs->regs[30]);
- if (gd->flags & GD_FLG_RELOC) {
printf("ELR: %lx\n", regs->elr - gd->reloc_off);printf("LR: %lx\n", regs->regs[30] - gd->reloc_off);- } else {
printf("ELR: %lx\n", regs->elr);printf("LR: %lx\n", regs->regs[30]);- } for (i = 0; i < 29; i += 2) printf("x%-2d: %016lx x%-2d: %016lx\n", i, regs->regs[i], i+1, regs->regs[i+1]);
Hi,
It is useful to show the relocated address, the kind of local mods I too have had for a while. But here you dropped the hw register values altogether, instead of displaying both, which I guess I am not the only one to not be happy about.
Regards, Karl
Hi,
On 22 January 2018 at 12:01, Karl Beldan karl.beldan@gmail.com wrote:
On Tue, Nov 28, 2017 at 10:08:08AM +0800, Peng Fan wrote:
After relocation, when error happends, it is hard to track ELR and LR with asm file objdumped from elf file.
So subtract the gd->reloc_off the reflect the compliation address.
Signed-off-by: Peng Fan peng.fan@nxp.com
arch/arm/lib/interrupts_64.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/arch/arm/lib/interrupts_64.c b/arch/arm/lib/interrupts_64.c index 7c9cfce69f..cbcfeec2b0 100644 --- a/arch/arm/lib/interrupts_64.c +++ b/arch/arm/lib/interrupts_64.c @@ -9,6 +9,7 @@ #include <linux/compiler.h> #include <efi_loader.h>
+DECLARE_GLOBAL_DATA_PTR;
int interrupt_init(void) { @@ -29,8 +30,13 @@ void show_regs(struct pt_regs *regs) { int i;
printf("ELR: %lx\n", regs->elr);printf("LR: %lx\n", regs->regs[30]);
if (gd->flags & GD_FLG_RELOC) {printf("ELR: %lx\n", regs->elr - gd->reloc_off);printf("LR: %lx\n", regs->regs[30] - gd->reloc_off);} else {printf("ELR: %lx\n", regs->elr);printf("LR: %lx\n", regs->regs[30]);} for (i = 0; i < 29; i += 2) printf("x%-2d: %016lx x%-2d: %016lx\n", i, regs->regs[i], i+1, regs->regs[i+1]);Hi,
It is useful to show the relocated address, the kind of local mods I too have had for a while. But here you dropped the hw register values altogether, instead of displaying both, which I guess I am not the only one to not be happy about.
Yes I agree that we should have both. Do you think you could do a patch?
Regards, Simon
On Sun, Feb 4, 2018 at 1:39 PM, Simon Glass sjg@chromium.org wrote:
Hi,
On 22 January 2018 at 12:01, Karl Beldan karl.beldan@gmail.com wrote:
On Tue, Nov 28, 2017 at 10:08:08AM +0800, Peng Fan wrote:
After relocation, when error happends, it is hard to track ELR and LR with asm file objdumped from elf file.
So subtract the gd->reloc_off the reflect the compliation address.
Signed-off-by: Peng Fan peng.fan@nxp.com
arch/arm/lib/interrupts_64.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/arch/arm/lib/interrupts_64.c b/arch/arm/lib/interrupts_64.c index 7c9cfce69f..cbcfeec2b0 100644 --- a/arch/arm/lib/interrupts_64.c +++ b/arch/arm/lib/interrupts_64.c @@ -9,6 +9,7 @@ #include <linux/compiler.h> #include <efi_loader.h>
+DECLARE_GLOBAL_DATA_PTR;
int interrupt_init(void) { @@ -29,8 +30,13 @@ void show_regs(struct pt_regs *regs) { int i;
printf("ELR: %lx\n", regs->elr);printf("LR: %lx\n", regs->regs[30]);
if (gd->flags & GD_FLG_RELOC) {printf("ELR: %lx\n", regs->elr - gd->reloc_off);printf("LR: %lx\n", regs->regs[30] - gd->reloc_off);} else {printf("ELR: %lx\n", regs->elr);printf("LR: %lx\n", regs->regs[30]);} for (i = 0; i < 29; i += 2) printf("x%-2d: %016lx x%-2d: %016lx\n", i, regs->regs[i], i+1, regs->regs[i+1]);Hi,
It is useful to show the relocated address, the kind of local mods I too have had for a while. But here you dropped the hw register values altogether, instead of displaying both, which I guess I am not the only one to not be happy about.
Yes I agree that we should have both. Do you think you could do a patch?
Hi,
Peng did reply with a patch[1], I was bugged by the formatting but no follow-up ensued, so unclear whether it felt annoying on anybody's side. Anyways sure, if the thread stalls for a few more days, I'll make sure to send a patch.
[1] https://marc.info/?l=u-boot&m=151674072412633&w=2
Regards, Karl
On Tue, Feb 06, 2018 at 12:40:39AM +0000, Karl Beldan wrote:
On Sun, Feb 4, 2018 at 1:39 PM, Simon Glass sjg@chromium.org wrote:
Hi,
On 22 January 2018 at 12:01, Karl Beldan karl.beldan@gmail.com wrote:
On Tue, Nov 28, 2017 at 10:08:08AM +0800, Peng Fan wrote:
After relocation, when error happends, it is hard to track ELR and LR with asm file objdumped from elf file.
So subtract the gd->reloc_off the reflect the compliation address.
Signed-off-by: Peng Fan peng.fan@nxp.com
arch/arm/lib/interrupts_64.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/arch/arm/lib/interrupts_64.c b/arch/arm/lib/interrupts_64.c index 7c9cfce69f..cbcfeec2b0 100644 --- a/arch/arm/lib/interrupts_64.c +++ b/arch/arm/lib/interrupts_64.c @@ -9,6 +9,7 @@ #include <linux/compiler.h> #include <efi_loader.h>
+DECLARE_GLOBAL_DATA_PTR;
int interrupt_init(void) { @@ -29,8 +30,13 @@ void show_regs(struct pt_regs *regs) { int i;
printf("ELR: %lx\n", regs->elr);printf("LR: %lx\n", regs->regs[30]);
if (gd->flags & GD_FLG_RELOC) {printf("ELR: %lx\n", regs->elr - gd->reloc_off);printf("LR: %lx\n", regs->regs[30] - gd->reloc_off);} else {printf("ELR: %lx\n", regs->elr);printf("LR: %lx\n", regs->regs[30]);} for (i = 0; i < 29; i += 2) printf("x%-2d: %016lx x%-2d: %016lx\n", i, regs->regs[i], i+1, regs->regs[i+1]);Hi,
It is useful to show the relocated address, the kind of local mods I too have had for a while. But here you dropped the hw register values altogether, instead of displaying both, which I guess I am not the only one to not be happy about.
Yes I agree that we should have both. Do you think you could do a patch?
Hi,
Peng did reply with a patch[1], I was bugged by the formatting but no follow-up ensued, so unclear whether it felt annoying on anybody's side. Anyways sure, if the thread stalls for a few more days, I'll make sure to send a patch.
For the record, I took your objects as "Changes still Requested" to the patch and I think/hope in patchwork I marked it as such.
participants (4)
-
Karl Beldan -
Peng Fan -
Simon Glass -
Tom Rini