Skip to contentSkip to conversation

s390x: Snippet fixes

PATCH Superseded Janosch Frank 4 participants 19 messages Nov 23, 2022 – Nov 30, 2022

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover s390x: Snippet fixes

A small set of fixes mostly related to the snippet support and error management.

Janosch Frank (5):
s390x: Add a linker script to assembly snippets
s390x: snippets: Fix SET_PSW_NEW_ADDR macro
lib: s390x: sie: Set guest memory pointer
s390x: Clear first stack frame and end backtrace early
lib: s390x: Handle debug prints for SIE exceptions correctly

 lib/s390x/interrupt.c       | 46 +++++++++++++++++++++++++++++++++----
 lib/s390x/sie.c             |  1 +
 lib/s390x/sie.h             |  2 ++
 lib/s390x/snippet.h         |  3 +--
 lib/s390x/stack.c           |  2 ++
 s390x/Makefile              |  5 ++--
 s390x/cpu.S                 |  6 +++--
 s390x/cstart64.S            |  2 ++
 s390x/mvpg-sie.c            |  2 +-
 s390x/pv-diags.c            |  6 ++---
 s390x/snippets/asm/flat.lds | 43 ++++++++++++++++++++++++++++++++++
 s390x/snippets/asm/macros.S |  2 +-
 12 files changed, 104 insertions(+), 16 deletions(-)
 create mode 100644 s390x/snippets/asm/flat.lds
Signature · 1 line
2.34.1
#1 lore raw

PATCH 1/5 s390x: Add a linker script to assembly snippets

A linker script has a few benefits:
- Random data doesn't end up in the binary breaking tests
- We can easily define a lowcore and load the snippet from 0x0 instead of 0x4000 which makes asm snippets behave like c snippets
- We can easily define an invalid PGM new PSW to ensure an exit on a guest PGM

---
 lib/s390x/snippet.h         |  3 +--
 s390x/Makefile              |  5 +++--
 s390x/mvpg-sie.c            |  2 +-
 s390x/pv-diags.c            |  6 +++---
 s390x/snippets/asm/flat.lds | 43 +++++++++++++++++++++++++++++++++++++
 5 files changed, 51 insertions(+), 8 deletions(-)
 create mode 100644 s390x/snippets/asm/flat.lds
lib/s390x/snippet.h +1 -2
index b17b2a4c..57045994 100644
@@ -32,8 +32,7 @@
 
 #define SNIPPET_PV_TWEAK0	0x42UL
 #define SNIPPET_PV_TWEAK1	0UL
-#define SNIPPET_OFF_C		0
-#define SNIPPET_OFF_ASM		0x4000
+#define SNIPPET_UNPACK_OFF	0
 
 
 /*
s390x/Makefile +3 -2
index bf1504f9..bb0f9eb8 100644
@@ -135,7 +135,8 @@ $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets)
 	$(CC) $(CFLAGS) -c -nostdlib -o $@ $<
 
 $(SNIPPET_DIR)/asm/%.gbin: $(SNIPPET_DIR)/asm/%.o
-	$(OBJCOPY) -O binary -j ".rodata" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $(patsubst %.gbin,%.o,$@) $@
+	$(CC) $(LDFLAGS) -o $@ -T $(SRCDIR)/s390x/snippets/asm/flat.lds $(patsubst %.gbin,%.o,$@)
+	$(OBJCOPY) -O binary -j ".rodata" -j ".lowcore" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $@ $@
 	truncate -s '%4096' $@
 
 $(SNIPPET_DIR)/c/%.gbin: $(SNIPPET_DIR)/c/%.o $(snippet_lib) $(FLATLIBS)
@@ -144,7 +145,7 @@ $(SNIPPET_DIR)/c/%.gbin: $(SNIPPET_DIR)/c/%.o $(snippet_lib) $(FLATLIBS)
 	truncate -s '%4096' $@
 
 $(SNIPPET_DIR)/asm/%.hdr: $(SNIPPET_DIR)/asm/%.gbin $(HOST_KEY_DOCUMENT)
-	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x4000,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
+	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x0,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
 
 $(SNIPPET_DIR)/c/%.hdr: $(SNIPPET_DIR)/c/%.gbin $(HOST_KEY_DOCUMENT)
 	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x0,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
s390x/mvpg-sie.c +1 -1
index 46a2edb6..99f4859b 100644
@@ -87,7 +87,7 @@ static void setup_guest(void)
 
 	snippet_setup_guest(&vm, false);
 	snippet_init(&vm, SNIPPET_NAME_START(c, mvpg_snippet),
-		     SNIPPET_LEN(c, mvpg_snippet), SNIPPET_OFF_C);
+		     SNIPPET_LEN(c, mvpg_snippet), SNIPPET_UNPACK_OFF);
 
 	/* Enable MVPG interpretation as we want to test KVM and not ourselves */
 	vm.sblk->eca = ECA_MVPGI;
s390x/pv-diags.c +3 -3
index 9ced68c7..5165937a 100644
@@ -28,7 +28,7 @@ static void test_diag_500(void)
 
 	snippet_pv_init(&vm, SNIPPET_NAME_START(asm, snippet_pv_diag_500),
 			SNIPPET_HDR_START(asm, snippet_pv_diag_500),
-			size_gbin, size_hdr, SNIPPET_OFF_ASM);
+			size_gbin, size_hdr, SNIPPET_UNPACK_OFF);
 
 	sie(&vm);
 	report(vm.sblk->icptcode == ICPT_PV_INSTR && vm.sblk->ipa == 0x8302 &&
@@ -83,7 +83,7 @@ static void test_diag_288(void)
 
 	snippet_pv_init(&vm, SNIPPET_NAME_START(asm, snippet_pv_diag_288),
 			SNIPPET_HDR_START(asm, snippet_pv_diag_288),
-			size_gbin, size_hdr, SNIPPET_OFF_ASM);
+			size_gbin, size_hdr, SNIPPET_UNPACK_OFF);
 
 	sie(&vm);
 	report(vm.sblk->icptcode == ICPT_PV_INSTR && vm.sblk->ipa == 0x8302 &&
@@ -124,7 +124,7 @@ static void test_diag_yield(void)
 
 	snippet_pv_init(&vm, SNIPPET_NAME_START(asm, snippet_pv_diag_yield),
 			SNIPPET_HDR_START(asm, snippet_pv_diag_yield),
-			size_gbin, size_hdr, SNIPPET_OFF_ASM);
+			size_gbin, size_hdr, SNIPPET_UNPACK_OFF);
 
 	/* 0x44 */
 	report_prefix_push("0x44");
s390x/snippets/asm/flat.lds new file +43 -0

43 lines not on this page · raw message

Signature · 1 line
2.34.1
#2 lore raw

Re: [kvm-unit-tests PATCH 1/5] s390x: Add a linker script to assembly snippets

Quoted 8 lines · On Wed, 23 Nov 2022 08:46:52 +0000 Janosch Frank <[email protected]> wrote:

A linker script has a few benefits:
- Random data doesn't end up in the binary breaking tests
- We can easily define a lowcore and load the snippet from 0x0 instead of 0x4000 which makes asm snippets behave like c snippets
- We can easily define an invalid PGM new PSW to ensure an exit on a guest PGM

looks good in general, but I have a few questions

Quoted 28 lines
---
 lib/s390x/snippet.h         |  3 +--
 s390x/Makefile              |  5 +++--
 s390x/mvpg-sie.c            |  2 +-
 s390x/pv-diags.c            |  6 +++---
 s390x/snippets/asm/flat.lds | 43 +++++++++++++++++++++++++++++++++++++
 5 files changed, 51 insertions(+), 8 deletions(-)
 create mode 100644 s390x/snippets/asm/flat.lds
lib/s390x/snippet.h +1 -2
index b17b2a4c..57045994 100644
@@ -32,8 +32,7 @@
 
 #define SNIPPET_PV_TWEAK0	0x42UL
 #define SNIPPET_PV_TWEAK1	0UL
-#define SNIPPET_OFF_C		0
-#define SNIPPET_OFF_ASM		0x4000
+#define SNIPPET_UNPACK_OFF	0
 
 
 /*
s390x/Makefile +0 -0
index bf1504f9..bb0f9eb8 100644

s390x/Makefile $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets) · line 138

Quoted 5 lines
s390x/Makefile +1 -1
@@ -135,7 +135,8 @@ $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets)
 	$(CC) $(CFLAGS) -c -nostdlib -o $@ $<
 
 $(SNIPPET_DIR)/asm/%.gbin: $(SNIPPET_DIR)/asm/%.o
-	$(OBJCOPY) -O binary -j ".rodata" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $(patsubst %.gbin,%.o,$@) $@
+	$(CC) $(LDFLAGS) -o $@ -T $(SRCDIR)/s390x/snippets/asm/flat.lds $(patsubst %.gbin,%.o,$@)

I think you can simply use $< instead of the patsubst expression

Quoted 82 lines
s390x/Makefile +2 -1
+	$(OBJCOPY) -O binary -j ".rodata" -j ".lowcore" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $@ $@
 	truncate -s '%4096' $@
 
 $(SNIPPET_DIR)/c/%.gbin: $(SNIPPET_DIR)/c/%.o $(snippet_lib) $(FLATLIBS)
@@ -144,7 +145,7 @@ $(SNIPPET_DIR)/c/%.gbin: $(SNIPPET_DIR)/c/%.o $(snippet_lib) $(FLATLIBS)
 	truncate -s '%4096' $@
 
 $(SNIPPET_DIR)/asm/%.hdr: $(SNIPPET_DIR)/asm/%.gbin $(HOST_KEY_DOCUMENT)
-	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x4000,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
+	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x0,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
 
 $(SNIPPET_DIR)/c/%.hdr: $(SNIPPET_DIR)/c/%.gbin $(HOST_KEY_DOCUMENT)
 	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x0,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
s390x/mvpg-sie.c +1 -1
index 46a2edb6..99f4859b 100644
@@ -87,7 +87,7 @@ static void setup_guest(void)
 
 	snippet_setup_guest(&vm, false);
 	snippet_init(&vm, SNIPPET_NAME_START(c, mvpg_snippet),
-		     SNIPPET_LEN(c, mvpg_snippet), SNIPPET_OFF_C);
+		     SNIPPET_LEN(c, mvpg_snippet), SNIPPET_UNPACK_OFF);
 
 	/* Enable MVPG interpretation as we want to test KVM and not ourselves */
 	vm.sblk->eca = ECA_MVPGI;
s390x/pv-diags.c +3 -3
index 9ced68c7..5165937a 100644
@@ -28,7 +28,7 @@ static void test_diag_500(void)
 
 	snippet_pv_init(&vm, SNIPPET_NAME_START(asm, snippet_pv_diag_500),
 			SNIPPET_HDR_START(asm, snippet_pv_diag_500),
-			size_gbin, size_hdr, SNIPPET_OFF_ASM);
+			size_gbin, size_hdr, SNIPPET_UNPACK_OFF);
 
 	sie(&vm);
 	report(vm.sblk->icptcode == ICPT_PV_INSTR && vm.sblk->ipa == 0x8302 &&
@@ -83,7 +83,7 @@ static void test_diag_288(void)
 
 	snippet_pv_init(&vm, SNIPPET_NAME_START(asm, snippet_pv_diag_288),
 			SNIPPET_HDR_START(asm, snippet_pv_diag_288),
-			size_gbin, size_hdr, SNIPPET_OFF_ASM);
+			size_gbin, size_hdr, SNIPPET_UNPACK_OFF);
 
 	sie(&vm);
 	report(vm.sblk->icptcode == ICPT_PV_INSTR && vm.sblk->ipa == 0x8302 &&
@@ -124,7 +124,7 @@ static void test_diag_yield(void)
 
 	snippet_pv_init(&vm, SNIPPET_NAME_START(asm, snippet_pv_diag_yield),
 			SNIPPET_HDR_START(asm, snippet_pv_diag_yield),
-			size_gbin, size_hdr, SNIPPET_OFF_ASM);
+			size_gbin, size_hdr, SNIPPET_UNPACK_OFF);
 
 	/* 0x44 */
 	report_prefix_push("0x44");
s390x/snippets/asm/flat.lds new file +19 -0
new file mode 100644
index 00000000..366d2d78
@@ -0,0 +1,43 @@
+SECTIONS
+{
+	.lowcore : {
+		/*
+		 * Initial short psw for disk boot, with 31 bit addressing for
+		 * non z/Arch environment compatibility and the instruction
+		 * address 0x4000.
+		 */
+		. = 0;
+		 LONG(0x00080000)
+		 LONG(0x80004000)
+		 /* Restart new PSW for booting via PSW restart. */
+		 . = 0x1a0;
+		 QUAD(0x0000000180000000)
+		 QUAD(0x0000000000004000)
+		 /*
+		  * Invalid PGM new PSW so we hopefully get a code 8
+		  * intercept on a PGM
+		  */

s390x/snippets/asm/flat.lds line 29

Quoted 10 lines
s390x/snippets/asm/flat.lds new file +10 -0
+		 . = 0x1d0;
+		 QUAD(0x0008000000000000)
+		 QUAD(0x0000000000000001)
+	}
+	. = 0x4000;
+	.text : {
+		*(.text)
+		*(.text.*)
+	}
+	. = ALIGN(64K);

any reason to align to 64k? (instead of e.g. 4k)

s390x/snippets/asm/flat.lds line 31

Quoted 2 lines
s390x/snippets/asm/flat.lds new file +2 -0
+	etext = .;
+	. = ALIGN(16);

do you need the ALIGN? I would think we are already aligned here

s390x/snippets/asm/flat.lds line 42

Quoted 11 lines
s390x/snippets/asm/flat.lds new file +11 -0
+	.data : {
+		*(.data)
+		*(.data.rel*)
+	}
+	. = ALIGN(16);
+	.rodata : { *(.rodata) *(.rodata.*) }
+	. = ALIGN(16);
+	__bss_start = .;
+	.bss : { *(.bss) }
+	__bss_end = .;
+	. = ALIGN(64K);

same question as above regarding 64k

Quoted 1 line
s390x/snippets/asm/flat.lds new file +1 -0
+}
#3 lore raw

Re: [kvm-unit-tests PATCH 1/5] s390x: Add a linker script to assembly snippets

Quoted 44 lines · Claudio Imbrenda
Quoted 8 lines · On Wed, 23 Nov 2022 08:46:52 +0000 Janosch Frank <[email protected]> wrote:

A linker script has a few benefits:
- Random data doesn't end up in the binary breaking tests
- We can easily define a lowcore and load the snippet from 0x0 instead of 0x4000 which makes asm snippets behave like c snippets
- We can easily define an invalid PGM new PSW to ensure an exit on a guest PGM

looks good in general, but I have a few questions

Quoted 30 lines
---
  lib/s390x/snippet.h         |  3 +--
  s390x/Makefile              |  5 +++--
  s390x/mvpg-sie.c            |  2 +-
  s390x/pv-diags.c            |  6 +++---
  s390x/snippets/asm/flat.lds | 43 +++++++++++++++++++++++++++++++++++++
  5 files changed, 51 insertions(+), 8 deletions(-)

create mode 100644 s390x/snippets/asm/flat.lds

lib/s390x/snippet.h +1 -2
index b17b2a4c..57045994 100644
@@ -32,8 +32,7 @@
 
  #define SNIPPET_PV_TWEAK0	0x42UL
  #define SNIPPET_PV_TWEAK1	0UL
-#define SNIPPET_OFF_C		0
-#define SNIPPET_OFF_ASM		0x4000
+#define SNIPPET_UNPACK_OFF	0
 
 
  /*
s390x/Makefile +0 -0
index bf1504f9..bb0f9eb8 100644
@@ -135,7 +135,8 @@ $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets)
  	$(CC) $(CFLAGS) -c -nostdlib -o $@ $<
 
Quoted 5 lines

s390x/Makefile $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets) · line 138

Quoted 3 lines
s390x/Makefile +1 -1
  $(SNIPPET_DIR)/asm/%.gbin: $(SNIPPET_DIR)/asm/%.o
-	$(OBJCOPY) -O binary -j ".rodata" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $(patsubst %.gbin,%.o,$@) $@
+	$(CC) $(LDFLAGS) -o $@ -T $(SRCDIR)/s390x/snippets/asm/flat.lds $(patsubst %.gbin,%.o,$@)

I think you can simply use $< instead of the patsubst expression

Right, I'll fix that momentarily.

[...]

Quoted 7 lines

s390x/Makefile $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets)

Quoted 5 lines
s390x/Makefile +5 -0
+	.text : {
+		*(.text)
+		*(.text.*)
+	}
+	. = ALIGN(64K);

any reason to align to 64k? (instead of e.g. 4k)

Well, I've copied that from s390x/flat.lds...
I'll have a look at the required alignments when I find time.

Quoted 22 lines

s390x/Makefile

Quoted 3 lines
s390x/Makefile +2 -0
+	etext = .;
+	. = ALIGN(16);

do you need the ALIGN? I would think we are already aligned here

s390x/Makefile

Quoted 11 lines
s390x/Makefile +11 -0
+	.data : {
+		*(.data)
+		*(.data.rel*)
+	}
+	. = ALIGN(16);
+	.rodata : { *(.rodata) *(.rodata.*) }
+	. = ALIGN(16);
+	__bss_start = .;
+	.bss : { *(.bss) }
+	__bss_end = .;
+	. = ALIGN(64K);

same question as above regarding 64k

Quoted 1 line

+}

#4 lore raw

Re: [kvm-unit-tests PATCH 1/5] s390x: Add a linker script to assembly snippets

In reply to Janosch Frank
Quoted 22 lines · Janosch Frank

A linker script has a few benefits:
- Random data doesn't end up in the binary breaking tests
- We can easily define a lowcore and load the snippet from 0x0 instead of 0x4000 which makes asm snippets behave like c snippets
- We can easily define an invalid PGM new PSW to ensure an exit on a guest PGM

---
 lib/s390x/snippet.h         |  3 +--
 s390x/Makefile              |  5 +++--
 s390x/mvpg-sie.c            |  2 +-
 s390x/pv-diags.c            |  6 +++---
 s390x/snippets/asm/flat.lds | 43 +++++++++++++++++++++++++++++++++++++
 5 files changed, 51 insertions(+), 8 deletions(-)
 create mode 100644 s390x/snippets/asm/flat.lds
lib/s390x/snippet.h +0 -0
index b17b2a4c..57045994 100644

lib/s390x/snippet.h line 35

Quoted 6 lines
lib/s390x/snippet.h +1 -2
@@ -32,8 +32,7 @@
 
 #define SNIPPET_PV_TWEAK0	0x42UL
 #define SNIPPET_PV_TWEAK1	0UL
-#define SNIPPET_OFF_C		0
-#define SNIPPET_OFF_ASM		0x4000
+#define SNIPPET_UNPACK_OFF	0

You could also get rid of the offset parameter, couldn't you?

Quoted 8 lines
lib/s390x/snippet.h +0 -0
 /*
s390x/Makefile +0 -0
index bf1504f9..bb0f9eb8 100644

s390x/Makefile $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets) · line 139

Quoted 6 lines
s390x/Makefile +2 -1
@@ -135,7 +135,8 @@ $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets)
 	$(CC) $(CFLAGS) -c -nostdlib -o $@ $<
 
 $(SNIPPET_DIR)/asm/%.gbin: $(SNIPPET_DIR)/asm/%.o
-	$(OBJCOPY) -O binary -j ".rodata" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $(patsubst %.gbin,%.o,$@) $@
+	$(CC) $(LDFLAGS) -o $@ -T $(SRCDIR)/s390x/snippets/asm/flat.lds $(patsubst %.gbin,%.o,$@)
+	$(OBJCOPY) -O binary -j ".rodata" -j ".lowcore" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $@ $@

I assume .bss=alloc allocates the bss in the binary...

Quoted 13 lines
s390x/Makefile +1 -1
 	truncate -s '%4096' $@
 
 $(SNIPPET_DIR)/c/%.gbin: $(SNIPPET_DIR)/c/%.o $(snippet_lib) $(FLATLIBS)
@@ -144,7 +145,7 @@ $(SNIPPET_DIR)/c/%.gbin: $(SNIPPET_DIR)/c/%.o $(snippet_lib) $(FLATLIBS)
 	truncate -s '%4096' $@
 
 $(SNIPPET_DIR)/asm/%.hdr: $(SNIPPET_DIR)/asm/%.gbin $(HOST_KEY_DOCUMENT)
-	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x4000,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
+	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x0,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
 
 $(SNIPPET_DIR)/c/%.hdr: $(SNIPPET_DIR)/c/%.gbin $(HOST_KEY_DOCUMENT)
 	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x0,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@

[...]

Quoted 35 lines
s390x/snippets/asm/flat.lds new file +29 -0
new file mode 100644
index 00000000..366d2d78
@@ -0,0 +1,43 @@
+SECTIONS
+{
+	.lowcore : {
+		/*
+		 * Initial short psw for disk boot, with 31 bit addressing for
+		 * non z/Arch environment compatibility and the instruction
+		 * address 0x4000.
+		 */
+		. = 0;
+		 LONG(0x00080000)
+		 LONG(0x80004000)
+		 /* Restart new PSW for booting via PSW restart. */
+		 . = 0x1a0;
+		 QUAD(0x0000000180000000)
+		 QUAD(0x0000000000004000)
+		 /*
+		  * Invalid PGM new PSW so we hopefully get a code 8
+		  * intercept on a PGM
+		  */
+		 . = 0x1d0;
+		 QUAD(0x0008000000000000)
+		 QUAD(0x0000000000000001)
+	}
+	. = 0x4000;
+	.text : {
+		*(.text)
+		*(.text.*)
+	}
+	. = ALIGN(64K);

s390x/snippets/asm/flat.lds line 39

Quoted 10 lines
s390x/snippets/asm/flat.lds new file +10 -0
+	etext = .;
+	. = ALIGN(16);
+	.data : {
+		*(.data)
+		*(.data.rel*)
+	}
+	. = ALIGN(16);
+	.rodata : { *(.rodata) *(.rodata.*) }
+	. = ALIGN(16);
+	__bss_start = .;

.. so the __bss symbols are not necessary.
But then, the c flat.lds has them too.

Quoted 4 lines
s390x/snippets/asm/flat.lds new file +4 -0
+	.bss : { *(.bss) }
+	__bss_end = .;
+	. = ALIGN(64K);
+}
#5 lore raw

Re: [kvm-unit-tests PATCH 1/5] s390x: Add a linker script to assembly snippets

Quoted 24 lines · Janis Schoetterl-Glausch
Quoted 23 lines · Janosch Frank

A linker script has a few benefits:
- Random data doesn't end up in the binary breaking tests
- We can easily define a lowcore and load the snippet from 0x0 instead of 0x4000 which makes asm snippets behave like c snippets
- We can easily define an invalid PGM new PSW to ensure an exit on a guest PGM

---
  lib/s390x/snippet.h         |  3 +--
  s390x/Makefile              |  5 +++--
  s390x/mvpg-sie.c            |  2 +-
  s390x/pv-diags.c            |  6 +++---
  s390x/snippets/asm/flat.lds | 43 +++++++++++++++++++++++++++++++++++++
  5 files changed, 51 insertions(+), 8 deletions(-)

create mode 100644 s390x/snippets/asm/flat.lds

lib/s390x/snippet.h +0 -0
index b17b2a4c..57045994 100644
@@ -32,8 +32,7 @@
 
Quoted 7 lines

lib/s390x/snippet.h line 35

Quoted 5 lines
lib/s390x/snippet.h +1 -2
  #define SNIPPET_PV_TWEAK0	0x42UL
  #define SNIPPET_PV_TWEAK1	0UL
-#define SNIPPET_OFF_C		0
-#define SNIPPET_OFF_ASM		0x4000
+#define SNIPPET_UNPACK_OFF	0

You could also get rid of the offset parameter, couldn't you?

Right

Quoted 10 lines
Quoted 10 lines
lib/s390x/snippet.h +0 -0
  /*
s390x/Makefile +0 -0
index bf1504f9..bb0f9eb8 100644
@@ -135,7 +135,8 @@ $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets)
  	$(CC) $(CFLAGS) -c -nostdlib -o $@ $<
 
Quoted 6 lines

s390x/Makefile $(SNIPPET_DIR)/c/%.o: $(SNIPPET_DIR)/c/%.c $(asm-offsets) · line 139

Quoted 4 lines
s390x/Makefile +2 -1
  $(SNIPPET_DIR)/asm/%.gbin: $(SNIPPET_DIR)/asm/%.o
-	$(OBJCOPY) -O binary -j ".rodata" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $(patsubst %.gbin,%.o,$@) $@
+	$(CC) $(LDFLAGS) -o $@ -T $(SRCDIR)/s390x/snippets/asm/flat.lds $(patsubst %.gbin,%.o,$@)
+	$(OBJCOPY) -O binary -j ".rodata" -j ".lowcore" -j ".text" -j ".data" -j ".bss" --set-section-flags .bss=alloc,load,contents $@ $@

I assume .bss=alloc allocates the bss in the binary...

And that's fine since I don't want to handle 0x3E PGMs/faults.

If bss is in the binary then it'll be made secure on initial image unpack, if it isn't, then we need a handler to import pages on a 0x3E. And I don't really want to do that since a 0x3E could also mean that we have an issue with the test or HW/FW.

Quoted 69 lines
Quoted 14 lines
s390x/Makefile +1 -1
  	truncate -s '%4096' $@
 
  $(SNIPPET_DIR)/c/%.gbin: $(SNIPPET_DIR)/c/%.o $(snippet_lib) $(FLATLIBS)
@@ -144,7 +145,7 @@ $(SNIPPET_DIR)/c/%.gbin: $(SNIPPET_DIR)/c/%.o $(snippet_lib) $(FLATLIBS)
  	truncate -s '%4096' $@
 
  $(SNIPPET_DIR)/asm/%.hdr: $(SNIPPET_DIR)/asm/%.gbin $(HOST_KEY_DOCUMENT)
-	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x4000,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
+	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x0,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@
 
  $(SNIPPET_DIR)/c/%.hdr: $(SNIPPET_DIR)/c/%.gbin $(HOST_KEY_DOCUMENT)
  	$(GEN_SE_HEADER) -k $(HOST_KEY_DOCUMENT) -c $<,0x0,0x00000000000000420000000000000000 --psw-addr 0x4000 -o $@

[...]

Quoted 35 lines
s390x/snippets/asm/flat.lds new file +29 -0
new file mode 100644
index 00000000..366d2d78
@@ -0,0 +1,43 @@
+SECTIONS
+{
+	.lowcore : {
+		/*
+		 * Initial short psw for disk boot, with 31 bit addressing for
+		 * non z/Arch environment compatibility and the instruction
+		 * address 0x4000.
+		 */
+		. = 0;
+		 LONG(0x00080000)
+		 LONG(0x80004000)
+		 /* Restart new PSW for booting via PSW restart. */
+		 . = 0x1a0;
+		 QUAD(0x0000000180000000)
+		 QUAD(0x0000000000004000)
+		 /*
+		  * Invalid PGM new PSW so we hopefully get a code 8
+		  * intercept on a PGM
+		  */
+		 . = 0x1d0;
+		 QUAD(0x0008000000000000)
+		 QUAD(0x0000000000000001)
+	}
+	. = 0x4000;
+	.text : {
+		*(.text)
+		*(.text.*)
+	}
+	. = ALIGN(64K);

s390x/snippets/asm/flat.lds line 39

Quoted 10 lines
s390x/snippets/asm/flat.lds new file +10 -0
+	etext = .;
+	. = ALIGN(16);
+	.data : {
+		*(.data)
+		*(.data.rel*)
+	}
+	. = ALIGN(16);
+	.rodata : { *(.rodata) *(.rodata.*) }
+	. = ALIGN(16);
+	__bss_start = .;

.. so the __bss symbols are not necessary.
But then, the c flat.lds has them too.

Quoted 4 lines
s390x/snippets/asm/flat.lds new file +4 -0
+	.bss : { *(.bss) }
+	__bss_end = .;
+	. = ALIGN(64K);
+}
#6 lore raw

PATCH 2/5 s390x: snippets: Fix SET_PSW_NEW_ADDR macro

In reply to Janosch Frank

Let's store the psw mask instead of the address of the location where we should load the mask from.

---
 s390x/snippets/asm/macros.S | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
s390x/snippets/asm/macros.S +1 -1
index 667fb6dc..09d7f5be 100644
@@ -18,7 +18,7 @@
  */
 .macro SET_PSW_NEW_ADDR reg, psw_new_addr, addr_psw
 larl	\reg, psw_mask_64
-stg	\reg, \addr_psw
+mvc	\addr_psw(8,%r0), 0(\reg)
 larl	\reg, \psw_new_addr
 stg	\reg, \addr_psw + 8
 .endm
Signature · 1 line
2.34.1
Reviewed-by on v1 2/5
#8 lore raw

Re: [kvm-unit-tests PATCH 2/5] s390x: snippets: Fix SET_PSW_NEW_ADDR macro

In reply to Janosch Frank
Quoted 4 lines · On Wed, 23 Nov 2022 08:46:53 +0000 Janosch Frank <[email protected]> wrote:

Let's store the psw mask instead of the address of the location where we should load the mask from.

Quoted 17 lines
---
 s390x/snippets/asm/macros.S | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
s390x/snippets/asm/macros.S +1 -1
index 667fb6dc..09d7f5be 100644
@@ -18,7 +18,7 @@
  */
 .macro SET_PSW_NEW_ADDR reg, psw_new_addr, addr_psw
 larl	\reg, psw_mask_64
-stg	\reg, \addr_psw
+mvc	\addr_psw(8,%r0), 0(\reg)
 larl	\reg, \psw_new_addr
 stg	\reg, \addr_psw + 8
 .endm
#9 lore raw

PATCH 3/5 lib: s390x: sie: Set guest memory pointer

In reply to Janosch Frank

Seems like it was introduced but never set. It's nicer to have a pointer than to cast the MSO of a VM.

---
 lib/s390x/sie.c | 1 +
 1 file changed, 1 insertion(+)
lib/s390x/sie.c +1 -0
index a71985b6..9241b4b4 100644
@@ -93,6 +93,7 @@ void sie_guest_create(struct vm *vm, uint64_t guest_mem, uint64_t guest_mem_len)
 
 	/* Guest memory chunks are always 1MB */
 	assert(!(guest_mem_len & ~HPAGE_MASK));
+	vm->guest_mem = (uint8_t *)guest_mem;
 	/* For non-PV guests we re-use the host's ASCE for ease of use */
 	vm->save_area.guest.asce = stctg(1);
 	/* Currently MSO/MSL is the easiest option */
Signature · 1 line
2.34.1
Reviewed-by on v1 3/5
#11 lore raw

Re: [kvm-unit-tests PATCH 3/5] lib: s390x: sie: Set guest memory pointer

In reply to Janosch Frank
Quoted 4 lines · On Wed, 23 Nov 2022 08:46:54 +0000 Janosch Frank <[email protected]> wrote:

Seems like it was introduced but never set. It's nicer to have a pointer than to cast the MSO of a VM.

Quoted 16 lines
---
 lib/s390x/sie.c | 1 +
 1 file changed, 1 insertion(+)
lib/s390x/sie.c +1 -0
index a71985b6..9241b4b4 100644
@@ -93,6 +93,7 @@ void sie_guest_create(struct vm *vm, uint64_t guest_mem, uint64_t guest_mem_len)
 
 	/* Guest memory chunks are always 1MB */
 	assert(!(guest_mem_len & ~HPAGE_MASK));
+	vm->guest_mem = (uint8_t *)guest_mem;
 	/* For non-PV guests we re-use the host's ASCE for ease of use */
 	vm->save_area.guest.asce = stctg(1);
 	/* Currently MSO/MSL is the easiest option */
#12 lore raw

PATCH 4/5 s390x: Clear first stack frame and end backtrace early

In reply to Janosch Frank

When setting the first stack frame to 0, we can check for a 0 backchain pointer when doing backtraces to know when to stop.

---
 lib/s390x/stack.c | 2 ++
 s390x/cstart64.S  | 2 ++
 2 files changed, 4 insertions(+)
lib/s390x/stack.c +2 -0
index e714e07c..9f234a12 100644
@@ -22,6 +22,8 @@ int backtrace_frame(const void *frame, const void **return_addrs, int max_depth)
 	for (depth = 0; stack && depth < max_depth; depth++) {
 		return_addrs[depth] = (void *)stack->grs[8];
 		stack = stack->back_chain;
+		if (!stack)
+			break;
 	}
 
 	return depth;
s390x/cstart64.S +2 -0
index 666a9567..6f83da2a 100644
@@ -37,6 +37,8 @@ start:
 	sam64				# Set addressing mode to 64 bit
 	/* setup stack */
 	larl	%r15, stackptr
+	/* Clear first stack frame */
+	xc      0(160,%r15), 0(%r15)
 	/* setup initial PSW mask + control registers*/
 	larl	%r1, initial_psw
 	lpswe	0(%r1)
Signature · 1 line
2.34.1
Reviewed-by on v1 4/5
#13 lore raw

Re: [kvm-unit-tests PATCH 4/5] s390x: Clear first stack frame and end backtrace early

Quoting Janosch Frank (2022-11-23 09:46:55)

Quoted 4 lines

When setting the first stack frame to 0, we can check for a 0 backchain pointer when doing backtraces to know when to stop.

Reviewed-by on v1 4/5
#14 lore raw

Re: [kvm-unit-tests PATCH 4/5] s390x: Clear first stack frame and end backtrace early

In reply to Janosch Frank
Quoted 4 lines · On Wed, 23 Nov 2022 08:46:55 +0000 Janosch Frank <[email protected]> wrote:

When setting the first stack frame to 0, we can check for a 0 backchain pointer when doing backtraces to know when to stop.

Quoted 31 lines
---
 lib/s390x/stack.c | 2 ++
 s390x/cstart64.S  | 2 ++
 2 files changed, 4 insertions(+)
lib/s390x/stack.c +2 -0
index e714e07c..9f234a12 100644
@@ -22,6 +22,8 @@ int backtrace_frame(const void *frame, const void **return_addrs, int max_depth)
 	for (depth = 0; stack && depth < max_depth; depth++) {
 		return_addrs[depth] = (void *)stack->grs[8];
 		stack = stack->back_chain;
+		if (!stack)
+			break;
 	}
 
 	return depth;
s390x/cstart64.S +2 -0
index 666a9567..6f83da2a 100644
@@ -37,6 +37,8 @@ start:
 	sam64				# Set addressing mode to 64 bit
 	/* setup stack */
 	larl	%r15, stackptr
+	/* Clear first stack frame */
+	xc      0(160,%r15), 0(%r15)
 	/* setup initial PSW mask + control registers*/
 	larl	%r1, initial_psw
 	lpswe	0(%r1)
#15 lore raw

PATCH 5/5 lib: s390x: Handle debug prints for SIE exceptions correctly

In reply to Janosch Frank

When we leave SIE due to an exception, we'll still have guest values in registers 0 - 13 and that's not clearly portraied in our debug prints. So let's fix that.

---
 lib/s390x/interrupt.c | 46 ++++++++++++++++++++++++++++++++++++++-----
 lib/s390x/sie.h       |  2 ++
 s390x/cpu.S           |  6 ++++--
 3 files changed, 47 insertions(+), 7 deletions(-)
lib/s390x/interrupt.c +41 -5
index dadb7415..ff47c2c2 100644
@@ -9,6 +9,7 @@
  */
 #include <libcflat.h>
 #include <asm/barrier.h>
+#include <asm/asm-offsets.h>
 #include <sclp.h>
 #include <interrupt.h>
 #include <sie.h>
@@ -188,9 +189,12 @@ static void print_storage_exception_information(void)
 	}
 }
 
-static void print_int_regs(struct stack_frame_int *stack)
+static void print_int_regs(struct stack_frame_int *stack, bool sie)
 {
+	struct kvm_s390_sie_block *sblk;
+
 	printf("\n");
+	printf("%s\n", sie ? "Guest registers:" : "Host registers:");
 	printf("GPRS:\n");
 	printf("%016lx %016lx %016lx %016lx\n",
 	       stack->grs1[0], stack->grs1[1], stack->grs0[0], stack->grs0[1]);
@@ -198,24 +202,56 @@ static void print_int_regs(struct stack_frame_int *stack)
 	       stack->grs0[2], stack->grs0[3], stack->grs0[4], stack->grs0[5]);
 	printf("%016lx %016lx %016lx %016lx\n",
 	       stack->grs0[6], stack->grs0[7], stack->grs0[8], stack->grs0[9]);
-	printf("%016lx %016lx %016lx %016lx\n",
-	       stack->grs0[10], stack->grs0[11], stack->grs0[12], stack->grs0[13]);
+
+	if (sie) {
+		sblk = (struct kvm_s390_sie_block *)stack->grs0[12];
+		printf("%016lx %016lx %016lx %016lx\n",
+		       stack->grs0[10], stack->grs0[11], sblk->gg14, sblk->gg15);
+	} else {
+		printf("%016lx %016lx %016lx %016lx\n",
+		       stack->grs0[10], stack->grs0[11], stack->grs0[12], stack->grs0[13]);
+	}
+
 	printf("\n");
 }
 
 static void print_pgm_info(struct stack_frame_int *stack)
 
 {
-	bool in_sie;
+	bool in_sie, in_sie_gregs;
+	struct vm_save_area *vregs;
 
 	in_sie = (lowcore.pgm_old_psw.addr >= (uintptr_t)sie_entry &&
 		  lowcore.pgm_old_psw.addr <= (uintptr_t)sie_exit);
+	in_sie_gregs = (lowcore.pgm_old_psw.addr >= (uintptr_t)sie_entry_gregs &&
+			lowcore.pgm_old_psw.addr <= (uintptr_t)sie_exit_gregs);
 
 	printf("\n");
 	printf("Unexpected program interrupt %s: %#x on cpu %d at %#lx, ilen %d\n",
 	       in_sie ? "in SIE" : "",
 	       lowcore.pgm_int_code, stap(), lowcore.pgm_old_psw.addr, lowcore.pgm_int_id);
-	print_int_regs(stack);
+
+	/*
+	 * If we fall out of SIE before loading the host registers,
+	 * then we need to do it here so we print the host registers
+	 * and not the guest registers.
+	 *
+	 * Back tracing is actually not a problem since SIE restores gr15.
+	 */
+	if (in_sie_gregs) {
+		print_int_regs(stack, true);
+		vregs = *((struct vm_save_area **)(stack->grs0[13] + __SF_SIE_SAVEAREA));
+
+		/*
+		 * The grs are not linear on the interrupt stack frame.
+		 * We copy 0 and 1 here and 2 - 15 with the memcopy below.
+		 */
+		stack->grs1[0] = vregs->host.grs[0];
+		stack->grs1[1] = vregs->host.grs[1];
+		/*  2 - 15 */
+		memcpy(stack->grs0, &vregs->host.grs[2], sizeof(stack->grs0) - 8);
+	}
+	print_int_regs(stack, false);
 	dump_stack();
 
 	/* Dump stack doesn't end with a \n so we add it here instead */
lib/s390x/sie.h +2 -0
index a27a8401..147cb0f2 100644
@@ -273,6 +273,8 @@ struct vm {
 
 extern void sie_entry(void);
 extern void sie_exit(void);
+extern void sie_entry_gregs(void);
+extern void sie_exit_gregs(void);
 extern void sie64a(struct kvm_s390_sie_block *sblk, struct vm_save_area *save_area);
 void sie(struct vm *vm);
 void sie_expect_validity(struct vm *vm);
s390x/cpu.S +4 -2
index 45bd551a..9155b044 100644
@@ -82,7 +82,8 @@ sie64a:
 	# Store scb and save_area pointer into stack frame
 	stg	%r2,__SF_SIE_CONTROL(%r15)	# save control block pointer
 	stg	%r3,__SF_SIE_SAVEAREA(%r15)	# save guest register save area
-
+.globl sie_entry_gregs
+sie_entry_gregs:
 	# Load guest's gprs, fprs and fpc
 	.irp i, 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15
 	ld	\i, \i * 8 + SIE_SAVEAREA_GUEST_FPRS(%r3)
@@ -121,7 +122,8 @@ sie_exit:
 	.endr
 	lfpc	SIE_SAVEAREA_HOST_FPC(%r14)
 	lmg	%r0,%r14,SIE_SAVEAREA_HOST_GRS(%r14)	# restore kernel registers
-
+.globl sie_exit_gregs
+sie_exit_gregs:
 	br	%r14
 
 	.align	8
Signature · 1 line
2.34.1
#16 lore raw

Re: [kvm-unit-tests PATCH 5/5] lib: s390x: Handle debug prints for SIE exceptions correctly

Quoted 3 lines · On Wed, 23 Nov 2022 08:46:56 +0000 Janosch Frank <[email protected]> wrote:

When we leave SIE due to an exception, we'll still have guest values in registers 0 - 13 and that's not clearly portraied in our debug prints. So let's fix that.

wouldn't it be cleaner to restore the registers in the interrupt handler? (I thought we were already doing it)

Quoted 132 lines
---
 lib/s390x/interrupt.c | 46 ++++++++++++++++++++++++++++++++++++++-----
 lib/s390x/sie.h       |  2 ++
 s390x/cpu.S           |  6 ++++--
 3 files changed, 47 insertions(+), 7 deletions(-)
lib/s390x/interrupt.c +41 -5
index dadb7415..ff47c2c2 100644
@@ -9,6 +9,7 @@
  */
 #include <libcflat.h>
 #include <asm/barrier.h>
+#include <asm/asm-offsets.h>
 #include <sclp.h>
 #include <interrupt.h>
 #include <sie.h>
@@ -188,9 +189,12 @@ static void print_storage_exception_information(void)
 	}
 }
 
-static void print_int_regs(struct stack_frame_int *stack)
+static void print_int_regs(struct stack_frame_int *stack, bool sie)
 {
+	struct kvm_s390_sie_block *sblk;
+
 	printf("\n");
+	printf("%s\n", sie ? "Guest registers:" : "Host registers:");
 	printf("GPRS:\n");
 	printf("%016lx %016lx %016lx %016lx\n",
 	       stack->grs1[0], stack->grs1[1], stack->grs0[0], stack->grs0[1]);
@@ -198,24 +202,56 @@ static void print_int_regs(struct stack_frame_int *stack)
 	       stack->grs0[2], stack->grs0[3], stack->grs0[4], stack->grs0[5]);
 	printf("%016lx %016lx %016lx %016lx\n",
 	       stack->grs0[6], stack->grs0[7], stack->grs0[8], stack->grs0[9]);
-	printf("%016lx %016lx %016lx %016lx\n",
-	       stack->grs0[10], stack->grs0[11], stack->grs0[12], stack->grs0[13]);
+
+	if (sie) {
+		sblk = (struct kvm_s390_sie_block *)stack->grs0[12];
+		printf("%016lx %016lx %016lx %016lx\n",
+		       stack->grs0[10], stack->grs0[11], sblk->gg14, sblk->gg15);
+	} else {
+		printf("%016lx %016lx %016lx %016lx\n",
+		       stack->grs0[10], stack->grs0[11], stack->grs0[12], stack->grs0[13]);
+	}
+
 	printf("\n");
 }
 
 static void print_pgm_info(struct stack_frame_int *stack)
 
 {
-	bool in_sie;
+	bool in_sie, in_sie_gregs;
+	struct vm_save_area *vregs;
 
 	in_sie = (lowcore.pgm_old_psw.addr >= (uintptr_t)sie_entry &&
 		  lowcore.pgm_old_psw.addr <= (uintptr_t)sie_exit);
+	in_sie_gregs = (lowcore.pgm_old_psw.addr >= (uintptr_t)sie_entry_gregs &&
+			lowcore.pgm_old_psw.addr <= (uintptr_t)sie_exit_gregs);
 
 	printf("\n");
 	printf("Unexpected program interrupt %s: %#x on cpu %d at %#lx, ilen %d\n",
 	       in_sie ? "in SIE" : "",
 	       lowcore.pgm_int_code, stap(), lowcore.pgm_old_psw.addr, lowcore.pgm_int_id);
-	print_int_regs(stack);
+
+	/*
+	 * If we fall out of SIE before loading the host registers,
+	 * then we need to do it here so we print the host registers
+	 * and not the guest registers.
+	 *
+	 * Back tracing is actually not a problem since SIE restores gr15.
+	 */
+	if (in_sie_gregs) {
+		print_int_regs(stack, true);
+		vregs = *((struct vm_save_area **)(stack->grs0[13] + __SF_SIE_SAVEAREA));
+
+		/*
+		 * The grs are not linear on the interrupt stack frame.
+		 * We copy 0 and 1 here and 2 - 15 with the memcopy below.
+		 */
+		stack->grs1[0] = vregs->host.grs[0];
+		stack->grs1[1] = vregs->host.grs[1];
+		/*  2 - 15 */
+		memcpy(stack->grs0, &vregs->host.grs[2], sizeof(stack->grs0) - 8);
+	}
+	print_int_regs(stack, false);
 	dump_stack();
 
 	/* Dump stack doesn't end with a \n so we add it here instead */
lib/s390x/sie.h +2 -0
index a27a8401..147cb0f2 100644
@@ -273,6 +273,8 @@ struct vm {
 
 extern void sie_entry(void);
 extern void sie_exit(void);
+extern void sie_entry_gregs(void);
+extern void sie_exit_gregs(void);
 extern void sie64a(struct kvm_s390_sie_block *sblk, struct vm_save_area *save_area);
 void sie(struct vm *vm);
 void sie_expect_validity(struct vm *vm);
s390x/cpu.S +4 -2
index 45bd551a..9155b044 100644
@@ -82,7 +82,8 @@ sie64a:
 	# Store scb and save_area pointer into stack frame
 	stg	%r2,__SF_SIE_CONTROL(%r15)	# save control block pointer
 	stg	%r3,__SF_SIE_SAVEAREA(%r15)	# save guest register save area
-
+.globl sie_entry_gregs
+sie_entry_gregs:
 	# Load guest's gprs, fprs and fpc
 	.irp i, 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15
 	ld	\i, \i * 8 + SIE_SAVEAREA_GUEST_FPRS(%r3)
@@ -121,7 +122,8 @@ sie_exit:
 	.endr
 	lfpc	SIE_SAVEAREA_HOST_FPC(%r14)
 	lmg	%r0,%r14,SIE_SAVEAREA_HOST_GRS(%r14)	# restore kernel registers
-
+.globl sie_exit_gregs
+sie_exit_gregs:
 	br	%r14
 
 	.align	8
#17 lore raw

Re: [kvm-unit-tests PATCH 5/5] lib: s390x: Handle debug prints for SIE exceptions correctly

Quoted 9 lines · Claudio Imbrenda
Quoted 3 lines · On Wed, 23 Nov 2022 08:46:56 +0000 Janosch Frank <[email protected]> wrote:

When we leave SIE due to an exception, we'll still have guest values in registers 0 - 13 and that's not clearly portraied in our debug prints. So let's fix that.

wouldn't it be cleaner to restore the registers in the interrupt handler? (I thought we were already doing it)

You mean RESTORE_REGS_STACK? Please don't make me write this in assembly...

RESTORE_REGS_STACK is doing test/pgm register swapping, it doesn't care if the test registers are "host" or "guest" registers.

#18 lore raw

Re: [kvm-unit-tests PATCH 5/5] lib: s390x: Handle debug prints for SIE exceptions correctly

Quoted 8 lines · On Wed, 30 Nov 2022 15:38:53 +0100 Janosch Frank <[email protected]> wrote:
Quoted 7 lines · Claudio Imbrenda
Quoted 3 lines · On Wed, 23 Nov 2022 08:46:56 +0000 Janosch Frank <[email protected]> wrote:

When we leave SIE due to an exception, we'll still have guest values in registers 0 - 13 and that's not clearly portraied in our debug prints. So let's fix that.

Quoted 7 lines
Quoted 2 lines

wouldn't it be cleaner to restore the registers in the interrupt handler? (I thought we were already doing it)

You mean RESTORE_REGS_STACK? Please don't make me write this in assembly...

RESTORE_REGS_STACK is doing test/pgm register swapping, it doesn't care if the test registers are "host" or "guest" registers.

fair enough :)