[PATCH] bpf: s390: add JIT support for bpf line info

[PATCH] bpf: s390: add JIT support for bpf line info

[Yauheni Kaliuta]

This adds support for generating bpf line info for JITed programs
like commit 6f20c71d8505 ("bpf: powerpc64: add JIT support for bpf
line info") does for powerpc.

Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
---

The patch is on top of "bpf: s390: add JIT support for multi-function
programs"

---
 arch/s390/net/bpf_jit_comp.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/s390/net/bpf_jit_comp.c b/arch/s390/net/bpf_jit_comp.c
index b6801d854c77..4ef783b67dfc 100644
--- a/arch/s390/net/bpf_jit_comp.c
+++ b/arch/s390/net/bpf_jit_comp.c
@@ -1420,6 +1420,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 	fp->jited_len = jit.size;
 
 	if (!fp->is_func || extra_pass) {
+		bpf_prog_fill_jited_linfo(fp, jit.addrs);
 free_addrs:
 		kfree(jit.addrs);
 		kfree(jit_data);
-- 
2.22.0

Re: [PATCH] bpf: s390: add JIT support for bpf line info

[Yauheni Kaliuta]

Hi!

Hold on, I should check it a bit more.

>>>>> On Thu, 29 Aug 2019 19:13:30 +0300, Yauheni Kaliuta  wrote:

 > This adds support for generating bpf line info for JITed programs
 > like commit 6f20c71d8505 ("bpf: powerpc64: add JIT support for bpf
 > line info") does for powerpc.

 > Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
 > ---

 > The patch is on top of "bpf: s390: add JIT support for multi-function
 > programs"

 > ---
 >  arch/s390/net/bpf_jit_comp.c | 1 +
 >  1 file changed, 1 insertion(+)

 > diff --git a/arch/s390/net/bpf_jit_comp.c b/arch/s390/net/bpf_jit_comp.c
 > index b6801d854c77..4ef783b67dfc 100644
 > --- a/arch/s390/net/bpf_jit_comp.c
 > +++ b/arch/s390/net/bpf_jit_comp.c
 > @@ -1420,6 +1420,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 > fp-> jited_len = jit.size;
 
 >  	if (!fp->is_func || extra_pass) {
 > +		bpf_prog_fill_jited_linfo(fp, jit.addrs);
 >  free_addrs:
 >  		kfree(jit.addrs);
 >  		kfree(jit_data);
 > -- 
 > 2.22.0


-- 
WBR,
Yauheni Kaliuta

[PATCH v2] bpf: s390: add JIT support for bpf line info

[Yauheni Kaliuta]

This adds support for generating bpf line info for JITed programs
like commit 6f20c71d8505 ("bpf: powerpc64: add JIT support for bpf
line info") does for powerpc, but it should pass the array starting
from 1 like x86, see commit 7c2e988f400e ("bpf: fix x64 JIT code
generation for jmp to 1st insn".

That fixes test_btf.

Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
---

The patch is on top of "bpf: s390: add JIT support for multi-function
programs"

V1->V1:

- pass address array starting from element 1.

---
 arch/s390/net/bpf_jit_comp.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/s390/net/bpf_jit_comp.c b/arch/s390/net/bpf_jit_comp.c
index b6801d854c77..ce88211b9c6c 100644
--- a/arch/s390/net/bpf_jit_comp.c
+++ b/arch/s390/net/bpf_jit_comp.c
@@ -1420,6 +1420,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 	fp->jited_len = jit.size;
 
 	if (!fp->is_func || extra_pass) {
+		bpf_prog_fill_jited_linfo(fp, jit.addrs + 1);
 free_addrs:
 		kfree(jit.addrs);
 		kfree(jit_data);
-- 
2.22.0

Re: [PATCH v2] bpf: s390: add JIT support for bpf line info

[Ilya Leoshkevich]

> Am 29.08.2019 um 22:02 schrieb Yauheni Kaliuta <yauheni.kaliuta@redhat.com>:
> 
> This adds support for generating bpf line info for JITed programs
> like commit 6f20c71d8505 ("bpf: powerpc64: add JIT support for bpf
> line info") does for powerpc, but it should pass the array starting
> from 1 like x86, see commit 7c2e988f400e ("bpf: fix x64 JIT code
> generation for jmp to 1st insn".
> 
> That fixes test_btf.
> 
> Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
> ---
> 
> The patch is on top of "bpf: s390: add JIT support for multi-function
> programs"
> 
> V1->V1:
> 
> - pass address array starting from element 1.
> 
> ---
> arch/s390/net/bpf_jit_comp.c | 1 +
> 1 file changed, 1 insertion(+)
> 
> diff --git a/arch/s390/net/bpf_jit_comp.c b/arch/s390/net/bpf_jit_comp.c
> index b6801d854c77..ce88211b9c6c 100644
> --- a/arch/s390/net/bpf_jit_comp.c
> +++ b/arch/s390/net/bpf_jit_comp.c
> @@ -1420,6 +1420,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
> 	fp->jited_len = jit.size;
> 
> 	if (!fp->is_func || extra_pass) {
> +		bpf_prog_fill_jited_linfo(fp, jit.addrs + 1);
> free_addrs:
> 		kfree(jit.addrs);
> 		kfree(jit_data);
> -- 
> 2.22.0
> 

Checkpatch complains about the missing ")" at the end of 7c2e988f400e
commit description. With that fixed:

Acked-by: Ilya Leoshkevich <iii@linux.ibm.com>
Tested-by: Ilya Leoshkevich <iii@linux.ibm.com>

Thanks!

Re: [PATCH v2] bpf: s390: add JIT support for bpf line info

[Yauheni Kaliuta]

Hi, Ilya!

>>>>> On Fri, 30 Aug 2019 13:39:19 +0200, Ilya Leoshkevich  wrote:

 >> Am 29.08.2019 um 22:02 schrieb Yauheni Kaliuta <yauheni.kaliuta@redhat.com>:
 >> 
 >> This adds support for generating bpf line info for JITed programs
 >> like commit 6f20c71d8505 ("bpf: powerpc64: add JIT support for bpf
 >> line info") does for powerpc, but it should pass the array starting
 >> from 1 like x86, see commit 7c2e988f400e ("bpf: fix x64 JIT code
 >> generation for jmp to 1st insn".
 >> 
 >> That fixes test_btf.
 >> 
 >> Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
 >> ---
 >> 
 >> The patch is on top of "bpf: s390: add JIT support for multi-function
 >> programs"
 >> 
 >> V1-> V1:
 >> 
 >> - pass address array starting from element 1.
 >> 
 >> ---
 >> arch/s390/net/bpf_jit_comp.c | 1 +
 >> 1 file changed, 1 insertion(+)
 >> 
 >> diff --git a/arch/s390/net/bpf_jit_comp.c b/arch/s390/net/bpf_jit_comp.c
 >> index b6801d854c77..ce88211b9c6c 100644
 >> --- a/arch/s390/net/bpf_jit_comp.c
 >> +++ b/arch/s390/net/bpf_jit_comp.c
 >> @@ -1420,6 +1420,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 fp-> jited_len = jit.size;
 >> 
 >> if (!fp->is_func || extra_pass) {
 >> +		bpf_prog_fill_jited_linfo(fp, jit.addrs + 1);
 >> free_addrs:
 >> kfree(jit.addrs);
 >> kfree(jit_data);
 >> -- 
 >> 2.22.0
 >> 

 > Checkpatch complains about the missing ")" at the end of
 > 7c2e988f400e commit description. With that fixed:

Huh, looks like I did not run checkpatch after the very last
editing :((

Thanks! Just a moment

 > Acked-by: Ilya Leoshkevich <iii@linux.ibm.com>
 > Tested-by: Ilya Leoshkevich <iii@linux.ibm.com>

 > Thanks!

-- 
WBR,
Yauheni Kaliuta

[PATCH v3] bpf: s390: add JIT support for bpf line info

[Yauheni Kaliuta]

This adds support for generating bpf line info for JITed programs
like commit 6f20c71d8505 ("bpf: powerpc64: add JIT support for bpf
line info") does for powerpc, but it should pass the array starting
from 1 like x86, see commit 7c2e988f400e ("bpf: fix x64 JIT code
generation for jmp to 1st insn").

That fixes test_btf.

Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
---

The patch is on top of "bpf: s390: add JIT support for multi-function
programs"
 
V1->V2:

- pass address array starting from element 1.

V2->V3:

- Fix braces in the commit message.

---
 arch/s390/net/bpf_jit_comp.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/s390/net/bpf_jit_comp.c b/arch/s390/net/bpf_jit_comp.c
index b6801d854c77..ce88211b9c6c 100644
--- a/arch/s390/net/bpf_jit_comp.c
+++ b/arch/s390/net/bpf_jit_comp.c
@@ -1420,6 +1420,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
 	fp->jited_len = jit.size;
 
 	if (!fp->is_func || extra_pass) {
+		bpf_prog_fill_jited_linfo(fp, jit.addrs + 1);
 free_addrs:
 		kfree(jit.addrs);
 		kfree(jit_data);
-- 
2.22.0

Re: [PATCH v3] bpf: s390: add JIT support for bpf line info

[Daniel Borkmann]

On 8/30/19 1:51 PM, Yauheni Kaliuta wrote:
> This adds support for generating bpf line info for JITed programs
> like commit 6f20c71d8505 ("bpf: powerpc64: add JIT support for bpf
> line info") does for powerpc, but it should pass the array starting
> from 1 like x86, see commit 7c2e988f400e ("bpf: fix x64 JIT code
> generation for jmp to 1st insn").
> 
> That fixes test_btf.
> 
> Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>

Applied, thanks! Few notes: please carry on Acked-by/Tested-by from
prior version since there haven't been fundamental changes. I've done
that manually now. Also, the reference above to x86 is somewhat
misleading since both JITs fill the addr array in a different manner
and the commit you've referenced is a bug fix (which would be good
if someone double checks on s390x).

Thanks,
Daniel