[PATCH v3 0/3] RISC-V: Introduce vSBI framework

[PATCH v3 0/3] RISC-V: Introduce vSBI framework

[Oleksii Kurochko]

Introduce the vSBI framework to handle extension ecall instruction calls
from a guest.

In this patch series, support is added for only a few extensions and FIDs,
just enough to boot the first guest domains and obtain logs from them. This
keeps the patch series independent from other ongoing work.

It was decided to start with support for the Legacy Extension, as it is still
supported by Linux (so we may need it anyway), and because some of its FIDs
require less functionality to implement at this stage compared to the more
modern extensions.

CI tests: https://gitlab.com/xen-project/people/olkur/xen/-/pipelines/2228014400

---
Changes in v3:
 - Address a comments from v2.
---
Changes in v2:
 - Address a comments from v1.
---

Oleksii Kurochko (3):
  xen/riscv: introduce vSBI extension framework
  xen/riscv: add RISC-V legacy SBI extension support for guests
  xen/riscv: add RISC-V virtual SBI base extension support for guests

 xen/arch/riscv/Makefile                |  1 +
 xen/arch/riscv/entry.S                 |  6 ++
 xen/arch/riscv/include/asm/processor.h |  1 +
 xen/arch/riscv/include/asm/sbi.h       | 21 ++++++-
 xen/arch/riscv/include/asm/vsbi.h      | 31 ++++++++++
 xen/arch/riscv/riscv64/asm-offsets.c   |  1 +
 xen/arch/riscv/setup.c                 |  3 +
 xen/arch/riscv/traps.c                 |  8 +++
 xen/arch/riscv/vsbi/Makefile           |  3 +
 xen/arch/riscv/vsbi/base-extension.c   | 78 ++++++++++++++++++++++++++
 xen/arch/riscv/vsbi/core.c             | 57 +++++++++++++++++++
 xen/arch/riscv/vsbi/legacy-extension.c | 64 +++++++++++++++++++++
 xen/arch/riscv/xen.lds.S               |  7 +++
 13 files changed, 279 insertions(+), 2 deletions(-)
 create mode 100644 xen/arch/riscv/include/asm/vsbi.h
 create mode 100644 xen/arch/riscv/vsbi/Makefile
 create mode 100644 xen/arch/riscv/vsbi/base-extension.c
 create mode 100644 xen/arch/riscv/vsbi/core.c
 create mode 100644 xen/arch/riscv/vsbi/legacy-extension.c

-- 
2.52.0

[PATCH v3 1/3] xen/riscv: introduce vSBI extension framework

[Oleksii Kurochko]

This commit introduces support for handling virtual SBI extensions in Xen.

The changes include:
- Added new vsbi/core.c and vsbi.h files to implement virtual SBI extension
  handling.
- Modified traps.c to handle CAUSE_VIRTUAL_SUPERVISOR_ECALL by calling
  vsbi_handle_ecall() when the trap originates from VS-mode.
- Updated xen.lds.S to include a new .vsbi.exts section for virtual SBI
  extension data.
- Updated Makefile to include the new vsbi/ directory in the build.
- Add hstatus register to struct cpu_user_regs as it is needed for
  a check that CAUSE_VIRTUAL_SUPERVISOR_ECALL happens from VS-mode.
  Also, add storing/restoring of hstatus register in handle_trap().
- Introduce vsbi_find_extension() to check if vsbi extension is supported
  by Xen. For now it is called only inside vsbi/core.c, but in future
  it is going to be called from other files.
- Introduce check_vsbi_ext_ranges() to check if there EIDs ranges
  overlapping between extensions.

The implementation allows for registration and handling of SBI
extensions via a new vsbi_ext structure and ".vsbi.exts" section,
enabling extensible virtual SBI support for RISC-V guests.

Note: All EIDs are printed in the format #%#lx and all FIDs in #%lu, as
the SBI spec uses these formats. Printing them this way makes it easier
to search for them in the SBI spec.

Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
---
Changes in v3:
 - Declare ext inside for() in vsbi_find_extension().
 - Use gprintk() instead of printk().
 - Drop `vcpu` argument for vsbi_handle_ecall() and
   ->handlers() as current is always used.
 - Correct commit message: s/#%#lu/#%lu.
---
Changes in v2:
 - s/struct regs/struct cpu_user_regs.
 - s/handle/handler.
 - Drop extid_ prefix inside VSBI_EXT_START().
 - use BUG_ON() instead of panic to be sure that 
   CAUSE_VIRTUAL_SUPERVISOR_ECALL comes from VS-mode.
 - s/ext_id/eid for vsbi_find_extension().
 - Add the comment above VSBI_EXT_START about [start,end] range.
 - Drop check "&& ext->handler" in vsbi_handle_ecall() as it isn't
   be so that handler() isn't provided.
 - s/vsbi.c/core.c
 - s/vsbi_ext/ext for local variable inside vsbi_find_extension().
 - Update the commit message: add a note about FID and EID printing formats,
   add some information about vsbo_find_extension() function, and add info
   about check_vsbi_ext_ranges().
 - Introduce check_vsbi_ext_ranges() to be sure that there is no overlapping
   in EIDs range(s).
 - Add storing/restoring of hstatus register in handle_trap()[entry.S].
---
 xen/arch/riscv/Makefile                |  1 +
 xen/arch/riscv/entry.S                 |  6 +++
 xen/arch/riscv/include/asm/processor.h |  1 +
 xen/arch/riscv/include/asm/vsbi.h      | 31 ++++++++++++++
 xen/arch/riscv/riscv64/asm-offsets.c   |  1 +
 xen/arch/riscv/setup.c                 |  3 ++
 xen/arch/riscv/traps.c                 |  8 ++++
 xen/arch/riscv/vsbi/Makefile           |  1 +
 xen/arch/riscv/vsbi/core.c             | 57 ++++++++++++++++++++++++++
 xen/arch/riscv/xen.lds.S               |  7 ++++
 10 files changed, 116 insertions(+)
 create mode 100644 xen/arch/riscv/include/asm/vsbi.h
 create mode 100644 xen/arch/riscv/vsbi/Makefile
 create mode 100644 xen/arch/riscv/vsbi/core.c

diff --git a/xen/arch/riscv/Makefile b/xen/arch/riscv/Makefile
index 9dde693db4..87c1148b00 100644
--- a/xen/arch/riscv/Makefile
+++ b/xen/arch/riscv/Makefile
@@ -20,6 +20,7 @@ obj-y += time.o
 obj-y += traps.o
 obj-y += vmid.o
 obj-y += vm_event.o
+obj-y += vsbi/
 
 $(TARGET): $(TARGET)-syms
 	$(OBJCOPY) -O binary -S $< $@
diff --git a/xen/arch/riscv/entry.S b/xen/arch/riscv/entry.S
index 4db818ba8d..202a35fb03 100644
--- a/xen/arch/riscv/entry.S
+++ b/xen/arch/riscv/entry.S
@@ -48,11 +48,17 @@ save_to_stack:
         csrr    t0, CSR_SSTATUS
         REG_S   t0, CPU_USER_REGS_SSTATUS(sp)
 
+        csrr    t0, CSR_HSTATUS
+        REG_S   t0, CPU_USER_REGS_HSTATUS(sp)
+
         mv      a0, sp
         call    do_trap
 
 restore_registers:
         /* Restore stack_cpu_regs */
+        REG_L	t0, CPU_USER_REGS_HSTATUS(sp)
+        csrw	CSR_HSTATUS, t0
+
         REG_L   t0, CPU_USER_REGS_SEPC(sp)
         csrw    CSR_SEPC, t0
         REG_L   t0, CPU_USER_REGS_SSTATUS(sp)
diff --git a/xen/arch/riscv/include/asm/processor.h b/xen/arch/riscv/include/asm/processor.h
index 2502045642..6b89df4a2d 100644
--- a/xen/arch/riscv/include/asm/processor.h
+++ b/xen/arch/riscv/include/asm/processor.h
@@ -49,6 +49,7 @@ struct cpu_user_regs
     unsigned long t6;
     unsigned long sepc;
     unsigned long sstatus;
+    unsigned long hstatus;
     /* pointer to previous stack_cpu_regs */
     unsigned long pregs;
 };
diff --git a/xen/arch/riscv/include/asm/vsbi.h b/xen/arch/riscv/include/asm/vsbi.h
new file mode 100644
index 0000000000..fa0cad604e
--- /dev/null
+++ b/xen/arch/riscv/include/asm/vsbi.h
@@ -0,0 +1,31 @@
+/* SPDX-License-Identifier:  GPL-2.0-only */
+
+#ifndef ASM_RISCV_VSBI_H
+#define ASM_RISCV_VSBI_H
+
+struct cpu_user_regs;
+
+struct vsbi_ext {
+    const char *name;
+    unsigned long eid_start;
+    unsigned long eid_end;
+    int (*handler)(unsigned long eid, unsigned long fid,
+                   struct cpu_user_regs *regs);
+};
+
+/* Ranges (start and end) are inclusive within an extension */
+#define VSBI_EXT(ext, start, end, handle)           \
+static const struct vsbi_ext vsbi_ext_##ext __used  \
+__section(".vsbi.exts") = {                         \
+    .name = #ext,                                   \
+    .eid_start = start,                             \
+    .eid_end = end,                                 \
+    .handler = handle,                              \
+};
+
+void vsbi_handle_ecall(struct cpu_user_regs *regs);
+const struct vsbi_ext *vsbi_find_extension(unsigned long eid);
+
+void check_vsbi_ext_ranges(void);
+
+#endif
diff --git a/xen/arch/riscv/riscv64/asm-offsets.c b/xen/arch/riscv/riscv64/asm-offsets.c
index 3b5daf3b36..472cced4f8 100644
--- a/xen/arch/riscv/riscv64/asm-offsets.c
+++ b/xen/arch/riscv/riscv64/asm-offsets.c
@@ -49,6 +49,7 @@ void asm_offsets(void)
     OFFSET(CPU_USER_REGS_T6, struct cpu_user_regs, t6);
     OFFSET(CPU_USER_REGS_SEPC, struct cpu_user_regs, sepc);
     OFFSET(CPU_USER_REGS_SSTATUS, struct cpu_user_regs, sstatus);
+    OFFSET(CPU_USER_REGS_HSTATUS, struct cpu_user_regs, hstatus);
     OFFSET(CPU_USER_REGS_PREGS, struct cpu_user_regs, pregs);
     BLANK();
     DEFINE(PCPU_INFO_SIZE, sizeof(struct pcpu_info));
diff --git a/xen/arch/riscv/setup.c b/xen/arch/riscv/setup.c
index 8f46f1a1de..9b4835960d 100644
--- a/xen/arch/riscv/setup.c
+++ b/xen/arch/riscv/setup.c
@@ -26,6 +26,7 @@
 #include <asm/sbi.h>
 #include <asm/setup.h>
 #include <asm/traps.h>
+#include <asm/vsbi.h>
 
 /* Xen stack for bringing up the first CPU. */
 unsigned char __initdata cpu0_boot_stack[STACK_SIZE]
@@ -110,6 +111,8 @@ void __init noreturn start_xen(unsigned long bootcpu_id,
 
     end_boot_allocator();
 
+    check_vsbi_ext_ranges();
+
     /*
      * The memory subsystem has been initialized, we can now switch from
      * early_boot -> boot.
diff --git a/xen/arch/riscv/traps.c b/xen/arch/riscv/traps.c
index f061004d83..e9c9677863 100644
--- a/xen/arch/riscv/traps.c
+++ b/xen/arch/riscv/traps.c
@@ -15,6 +15,7 @@
 #include <asm/processor.h>
 #include <asm/riscv_encoding.h>
 #include <asm/traps.h>
+#include <asm/vsbi.h>
 
 /*
  * Initialize the trap handling.
@@ -114,6 +115,13 @@ void do_trap(struct cpu_user_regs *cpu_regs)
 
     switch ( cause )
     {
+    case CAUSE_VIRTUAL_SUPERVISOR_ECALL:
+        /* CAUSE_VIRTUAL_SUPERVISOR_ECALL should come from VS-mode */
+        BUG_ON(!(cpu_regs->hstatus & HSTATUS_SPV));
+
+        vsbi_handle_ecall(cpu_regs);
+        break;
+
     case CAUSE_ILLEGAL_INSTRUCTION:
         if ( do_bug_frame(cpu_regs, pc) >= 0 )
         {
diff --git a/xen/arch/riscv/vsbi/Makefile b/xen/arch/riscv/vsbi/Makefile
new file mode 100644
index 0000000000..820eb10ac2
--- /dev/null
+++ b/xen/arch/riscv/vsbi/Makefile
@@ -0,0 +1 @@
+obj-y += core.o
diff --git a/xen/arch/riscv/vsbi/core.c b/xen/arch/riscv/vsbi/core.c
new file mode 100644
index 0000000000..f5ded8f676
--- /dev/null
+++ b/xen/arch/riscv/vsbi/core.c
@@ -0,0 +1,57 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#include <xen/sched.h>
+
+#include <asm/processor.h>
+#include <asm/sbi.h>
+#include <asm/vsbi.h>
+
+extern const struct vsbi_ext _svsbi_exts[], _evsbi_exts[];
+
+void __init check_vsbi_ext_ranges(void)
+{
+    for ( const struct vsbi_ext *a = _svsbi_exts; a != _evsbi_exts; a++ )
+        for ( const struct vsbi_ext *b = a + 1; b != _evsbi_exts; b++ )
+            if ( !(a->eid_end < b->eid_start || b->eid_end < a->eid_start) )
+                panic("EID range overlap detected: "
+                      "%s:[#%#lx..#%#lx] vs %s:[#%#lx..#%#lx]\n",
+                      a->name, a->eid_start, a->eid_end,
+                      b->name, b->eid_start, b->eid_end);
+}
+
+const struct vsbi_ext *vsbi_find_extension(unsigned long eid)
+{
+    for ( const struct vsbi_ext *ext = _svsbi_exts;
+          ext != _evsbi_exts;
+          ext++ )
+        if ( (eid >= ext->eid_start) && (eid <= ext->eid_end) )
+            return ext;
+
+    return NULL;
+}
+
+void vsbi_handle_ecall(struct cpu_user_regs *regs)
+{
+    const unsigned long eid = regs->a7;
+    const unsigned long fid = regs->a6;
+    const struct vsbi_ext *ext = vsbi_find_extension(eid);
+    int ret;
+
+    if ( ext )
+        ret = ext->handler(eid, fid, regs);
+    else
+    {
+        gprintk(XENLOG_ERR, "Unsupported Guest SBI EID #%#lx, FID #%lu\n",
+                eid, regs->a1);
+        ret = SBI_ERR_NOT_SUPPORTED;
+    }
+
+    /*
+     * The ecall instruction is not part of the RISC-V C extension (compressed
+     * instructions), so it is always 4 bytes long. Therefore, it is safe to
+     * use a fixed length of 4 bytes instead of reading guest memory to
+     * determine the instruction length.
+     */
+    regs->sepc += 4;
+    regs->a0 = ret;
+}
diff --git a/xen/arch/riscv/xen.lds.S b/xen/arch/riscv/xen.lds.S
index 45d2e053d0..331a7d63d3 100644
--- a/xen/arch/riscv/xen.lds.S
+++ b/xen/arch/riscv/xen.lds.S
@@ -61,6 +61,13 @@ SECTIONS
         __note_gnu_build_id_end = .;
     } :note :text
     #endif
+
+    . = ALIGN(POINTER_ALIGN);
+    DECL_SECTION(.vsbi.exts) {
+        _svsbi_exts = .;
+        *(.vsbi.exts)
+        _evsbi_exts = .;
+    } :text
     _erodata = .;                /* End of read-only data */
 
     . = ALIGN(PAGE_SIZE);
-- 
2.52.0

[PATCH v3 2/3] xen/riscv: add RISC-V legacy SBI extension support for guests

[Oleksii Kurochko]

This commit adds support for legacy SBI extensions (version 0.1) in Xen
for guest domains.

The changes include:
1. Define all legacy SBI extension IDs (0x0 to 0x8) for better clarity and
   completeness.
2. Implement handling of legacy SBI extensions, starting with support for
   SBI_EXT_0_1_CONSOLE_PUTCHAR. SBI_EXT_0_1_CONSOLE_GETCHAR is marked as
   not supported as legacy SBI console related stuff is expected to be used
   only for early debugging of guest.

The implementation uses the existing virtual SBI framework to handle legacy
SBI ecalls, ensuring compatibility with older SBI specifications in
RISC-V guests.

Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
---
Changes in v3:
 - s/vsbi_print_line/vsbi_print_char.
 - s/cd/currd in vsbi_print_char().
 - Drop a cast of regs->a0 when vsbi_print_line() is called.
 - Use current instead of `vcpu` argument as it was dropped from
   vsbi_base_ecall_handler() prototype.
---
Changes in v2:
 - s/vsbi-legacy-extension.*/legacy-extension.*
 - Correct padding for SBI_EXT_0_1_* macros.
 - Set ret = SBI_ERR_NOT_SUPPORTED; in the case of vsbi_legacy_ecall_handler()
   instead of regs->a0 = ... as regs->a0 will be overwritten in
   vsbi_handle_ecall().
 - Use domain_crash() instead of panic() in vsbi_legacy_ecall_handler() and
   add TODO.
 - Use newly introduced VSBI_EXT macros instead of VSBI_EXT_{START,END}.
 - Introduce vsbi_print_line() and use it instead of plain printk() in the
   handler of SBI_EXT_0_1_CONSOLE_PUTCHAR.
---
 xen/arch/riscv/include/asm/sbi.h       | 11 ++++-
 xen/arch/riscv/vsbi/Makefile           |  1 +
 xen/arch/riscv/vsbi/legacy-extension.c | 64 ++++++++++++++++++++++++++
 3 files changed, 74 insertions(+), 2 deletions(-)
 create mode 100644 xen/arch/riscv/vsbi/legacy-extension.c

diff --git a/xen/arch/riscv/include/asm/sbi.h b/xen/arch/riscv/include/asm/sbi.h
index ade24a572d..751bae6d66 100644
--- a/xen/arch/riscv/include/asm/sbi.h
+++ b/xen/arch/riscv/include/asm/sbi.h
@@ -14,8 +14,15 @@
 
 #include <xen/cpumask.h>
 
-#define SBI_EXT_0_1_CONSOLE_PUTCHAR		0x1
-#define SBI_EXT_0_1_SHUTDOWN			0x8
+#define SBI_EXT_0_1_SET_TIMER               0x0
+#define SBI_EXT_0_1_CONSOLE_PUTCHAR         0x1
+#define SBI_EXT_0_1_CONSOLE_GETCHAR         0x2
+#define SBI_EXT_0_1_CLEAR_IPI               0x3
+#define SBI_EXT_0_1_SEND_IPI                0x4
+#define SBI_EXT_0_1_REMOTE_FENCE_I          0x5
+#define SBI_EXT_0_1_REMOTE_SFENCE_VMA       0x6
+#define SBI_EXT_0_1_REMOTE_SFENCE_VMA_ASID  0x7
+#define SBI_EXT_0_1_SHUTDOWN                0x8
 
 #define SBI_EXT_BASE                    0x10
 #define SBI_EXT_RFENCE                  0x52464E43
diff --git a/xen/arch/riscv/vsbi/Makefile b/xen/arch/riscv/vsbi/Makefile
index 820eb10ac2..bc5755cb13 100644
--- a/xen/arch/riscv/vsbi/Makefile
+++ b/xen/arch/riscv/vsbi/Makefile
@@ -1 +1,2 @@
 obj-y += core.o
+obj-y += legacy-extension.o
diff --git a/xen/arch/riscv/vsbi/legacy-extension.c b/xen/arch/riscv/vsbi/legacy-extension.c
new file mode 100644
index 0000000000..ca869942d6
--- /dev/null
+++ b/xen/arch/riscv/vsbi/legacy-extension.c
@@ -0,0 +1,64 @@
+
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#include <xen/console.h>
+#include <xen/lib.h>
+#include <xen/sched.h>
+
+#include <asm/processor.h>
+#include <asm/vsbi.h>
+
+static void vsbi_print_char(char c)
+{
+    struct domain *currd = current->domain;
+    struct domain_console *cons = currd->console;
+
+    if ( !is_console_printable(c) )
+        return;
+
+    spin_lock(&cons->lock);
+    ASSERT(cons->idx < ARRAY_SIZE(cons->buf));
+    if ( c != '\n' )
+        cons->buf[cons->idx++] = c;
+    if ( (cons->idx == (ARRAY_SIZE(cons->buf) - 1)) || (c == '\n') )
+    {
+        cons->buf[cons->idx] = '\0';
+        guest_printk(currd, XENLOG_G_DEBUG "%s\n", cons->buf);
+        cons->idx = 0;
+    }
+    spin_unlock(&cons->lock);
+}
+
+static int vsbi_legacy_ecall_handler(unsigned long eid, unsigned long fid,
+                                     struct cpu_user_regs *regs)
+{
+    int ret = 0;
+
+    switch ( eid )
+    {
+    case SBI_EXT_0_1_CONSOLE_PUTCHAR:
+        vsbi_print_char(regs->a0);
+        break;
+
+    case SBI_EXT_0_1_CONSOLE_GETCHAR:
+        ret = SBI_ERR_NOT_SUPPORTED;
+        break;
+
+    default:
+        /*
+         * TODO: domain_crash() is acceptable here while things are still under
+         * development.
+         * It shouldn't stay like this in the end though: guests should not
+         * be punished like this for something Xen hasn't implemented.
+         */
+        domain_crash(current->domain,
+                     "%s: Unsupported ecall: FID: #%lx, EID: #%lx\n",
+                     __func__, fid, eid);
+        break;
+    }
+
+    return ret;
+}
+
+VSBI_EXT(legacy, SBI_EXT_0_1_SET_TIMER, SBI_EXT_0_1_SHUTDOWN,
+         vsbi_legacy_ecall_handler);
-- 
2.52.0

[PATCH v3 3/3] xen/riscv: add RISC-V virtual SBI base extension support for guests

[Oleksii Kurochko]

Add support of virtual SBI base extension calls for RISC-V guests, delegating
hardware-specific queries to the underlying SBI and handling version and
firmware ID queries directly.

The changes include:
1. Define new SBI base extension function IDs (SBI_EXT_BASE_GET_MVENDORID,
   SBI_EXT_BASE_GET_MARCHID, SBI_EXT_BASE_GET_MIMPID).
2. Introduce XEN_SBI_VER_MAJOR, XEN_SBI_VER_MINOR for imeplenataion of
   SBI_EXT_BASE_GET_SPEC_VERSION.
4. Introduce SBI_XEN_IMPID to implement SBI_EXT_BASE_GET_IMP_ID.
5. Implement handling of SBI base extension functions, including version,
   firmware ID, and machine-specific queries.

Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
---
Changes in v3:
 - s/XEN_SBI_IMPID/SBI_XEN_IMPID
 - Add ASSERT(eid == SBI_EXT_BASE) in vsbi_base_ecall_handler().
 - Fix code style for switch/case.
 - Use current instead of `vcpu` argument as it was dropped from
   vsbi_base_ecall_handler() prototype.
 - Add comments for define-s XEN_SBI_VER_{MAJOR, MINOR} and SBI_XEN_IMPID.
---
Changes in v2:
 - s/vsbi-base-extension.*/base-extension.*
 - Introduce VCPU_SBI_IMPID, XEN_SBI_VER_MINOR and XEN_SBI_VER_MAJOR.
 - Return VCPU_SBI_IMPID in the case of SBI_EXT_BASE_GET_IMP_ID.
 - Return Xen version in the case of SBI_EXT_BASE_GET_IMP_VERSION.
 - Use domain_crash() instead of panic() for default case.
 - For SBI_EXT_BASE_GET_{MVENDORID,MARCHID,MIMPID} abd SBI_EXT_BASE_PROBE_EXT
   add handling of a domain is h/w or not.
---
 xen/arch/riscv/include/asm/sbi.h     | 10 ++++
 xen/arch/riscv/vsbi/Makefile         |  1 +
 xen/arch/riscv/vsbi/base-extension.c | 78 ++++++++++++++++++++++++++++
 3 files changed, 89 insertions(+)
 create mode 100644 xen/arch/riscv/vsbi/base-extension.c

diff --git a/xen/arch/riscv/include/asm/sbi.h b/xen/arch/riscv/include/asm/sbi.h
index 751bae6d66..a88d3d5712 100644
--- a/xen/arch/riscv/include/asm/sbi.h
+++ b/xen/arch/riscv/include/asm/sbi.h
@@ -14,6 +14,13 @@
 
 #include <xen/cpumask.h>
 
+/* Xen-controlled SBI version reported to guests */
+#define XEN_SBI_VER_MAJOR 0
+#define XEN_SBI_VER_MINOR 2
+
+/* SBI-defined implementation ID */
+#define SBI_XEN_IMPID 7
+
 #define SBI_EXT_0_1_SET_TIMER               0x0
 #define SBI_EXT_0_1_CONSOLE_PUTCHAR         0x1
 #define SBI_EXT_0_1_CONSOLE_GETCHAR         0x2
@@ -32,6 +39,9 @@
 #define SBI_EXT_BASE_GET_IMP_ID         0x1
 #define SBI_EXT_BASE_GET_IMP_VERSION    0x2
 #define SBI_EXT_BASE_PROBE_EXT          0x3
+#define SBI_EXT_BASE_GET_MVENDORID      0x4
+#define SBI_EXT_BASE_GET_MARCHID        0x5
+#define SBI_EXT_BASE_GET_MIMPID         0x6
 
 /* SBI function IDs for RFENCE extension */
 #define SBI_EXT_RFENCE_REMOTE_FENCE_I           0x0
diff --git a/xen/arch/riscv/vsbi/Makefile b/xen/arch/riscv/vsbi/Makefile
index bc5755cb13..8ce470f787 100644
--- a/xen/arch/riscv/vsbi/Makefile
+++ b/xen/arch/riscv/vsbi/Makefile
@@ -1,2 +1,3 @@
+obj-y += base-extension.o
 obj-y += core.o
 obj-y += legacy-extension.o
diff --git a/xen/arch/riscv/vsbi/base-extension.c b/xen/arch/riscv/vsbi/base-extension.c
new file mode 100644
index 0000000000..0245ff630e
--- /dev/null
+++ b/xen/arch/riscv/vsbi/base-extension.c
@@ -0,0 +1,78 @@
+
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#include <xen/lib.h>
+#include <xen/sched.h>
+#include <xen/version.h>
+
+#include <asm/processor.h>
+#include <asm/sbi.h>
+#include <asm/vsbi.h>
+
+static int vsbi_base_ecall_handler(unsigned long eid, unsigned long fid,
+                                   struct cpu_user_regs *regs)
+{
+    int ret = 0;
+    struct sbiret sbi_ret;
+
+    ASSERT(eid == SBI_EXT_BASE);
+
+    switch ( fid )
+    {
+    case SBI_EXT_BASE_GET_SPEC_VERSION:
+        regs->a1 = MASK_INSR(XEN_SBI_VER_MAJOR, SBI_SPEC_VERSION_MAJOR_MASK) |
+                   XEN_SBI_VER_MINOR;
+        break;
+
+    case SBI_EXT_BASE_GET_IMP_ID:
+        regs->a1 = SBI_XEN_IMPID;
+        break;
+
+    case SBI_EXT_BASE_GET_IMP_VERSION:
+        regs->a1 = (xen_major_version() << 16) | xen_minor_version();
+        break;
+
+    case SBI_EXT_BASE_GET_MVENDORID:
+    case SBI_EXT_BASE_GET_MARCHID:
+    case SBI_EXT_BASE_GET_MIMPID:
+        if ( is_hardware_domain(current->domain) )
+        {
+            sbi_ret = sbi_ecall(SBI_EXT_BASE, fid, 0, 0, 0, 0, 0, 0);
+            ret = sbi_ret.error;
+            regs->a1 = sbi_ret.value;
+        }
+        else
+            /*
+             * vSBI should present a consistent, virtualized view to guests.
+             * In particular, DomU-visible data must remain stable across
+             * migration and must not expose hardware-specific details.
+             *
+             * These register(s) must be readable in any implementation,
+             * but a value of 0 can be returned to indicate the field
+             * is not implemented.
+             */
+            regs->a1 = 0;
+
+        break;
+
+    case SBI_EXT_BASE_PROBE_EXT:
+        regs->a1 = vsbi_find_extension(regs->a0) ? 1 : 0;
+        break;
+
+    default:
+        /*
+         * TODO: domain_crash() is acceptable here while things are still under
+         * development.
+         * It shouldn't stay like this in the end though: guests should not
+         * be punished like this for something Xen hasn't implemented.
+         */
+        domain_crash(current->domain,
+                     "%s: Unsupported ecall: FID: #%lx, EID: #%lx\n",
+                     __func__, fid, eid);
+        break;
+    }
+
+    return ret;
+}
+
+VSBI_EXT(base, SBI_EXT_BASE, SBI_EXT_BASE, vsbi_base_ecall_handler)
-- 
2.52.0

Re: [PATCH v3 1/3] xen/riscv: introduce vSBI extension framework

[Jan Beulich]

On 22.12.2025 17:37, Oleksii Kurochko wrote:
> This commit introduces support for handling virtual SBI extensions in Xen.
> 
> The changes include:
> - Added new vsbi/core.c and vsbi.h files to implement virtual SBI extension
>   handling.
> - Modified traps.c to handle CAUSE_VIRTUAL_SUPERVISOR_ECALL by calling
>   vsbi_handle_ecall() when the trap originates from VS-mode.
> - Updated xen.lds.S to include a new .vsbi.exts section for virtual SBI
>   extension data.
> - Updated Makefile to include the new vsbi/ directory in the build.
> - Add hstatus register to struct cpu_user_regs as it is needed for
>   a check that CAUSE_VIRTUAL_SUPERVISOR_ECALL happens from VS-mode.
>   Also, add storing/restoring of hstatus register in handle_trap().
> - Introduce vsbi_find_extension() to check if vsbi extension is supported
>   by Xen. For now it is called only inside vsbi/core.c, but in future
>   it is going to be called from other files.
> - Introduce check_vsbi_ext_ranges() to check if there EIDs ranges
>   overlapping between extensions.
> 
> The implementation allows for registration and handling of SBI
> extensions via a new vsbi_ext structure and ".vsbi.exts" section,
> enabling extensible virtual SBI support for RISC-V guests.
> 
> Note: All EIDs are printed in the format #%#lx and all FIDs in #%lu, as
> the SBI spec uses these formats. Printing them this way makes it easier
> to search for them in the SBI spec.
> 
> Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>

Acked-by: Jan Beulich <jbeulich@suse.com>

I'd like to note though that ...

> --- /dev/null
> +++ b/xen/arch/riscv/vsbi/core.c
> @@ -0,0 +1,57 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +
> +#include <xen/sched.h>
> +
> +#include <asm/processor.h>
> +#include <asm/sbi.h>
> +#include <asm/vsbi.h>
> +
> +extern const struct vsbi_ext _svsbi_exts[], _evsbi_exts[];
> +
> +void __init check_vsbi_ext_ranges(void)
> +{
> +    for ( const struct vsbi_ext *a = _svsbi_exts; a != _evsbi_exts; a++ )
> +        for ( const struct vsbi_ext *b = a + 1; b != _evsbi_exts; b++ )
> +            if ( !(a->eid_end < b->eid_start || b->eid_end < a->eid_start) )
> +                panic("EID range overlap detected: "
> +                      "%s:[#%#lx..#%#lx] vs %s:[#%#lx..#%#lx]\n",
> +                      a->name, a->eid_start, a->eid_end,
> +                      b->name, b->eid_start, b->eid_end);
> +}
> +
> +const struct vsbi_ext *vsbi_find_extension(unsigned long eid)
> +{
> +    for ( const struct vsbi_ext *ext = _svsbi_exts;
> +          ext != _evsbi_exts;
> +          ext++ )
> +        if ( (eid >= ext->eid_start) && (eid <= ext->eid_end) )
> +            return ext;
> +
> +    return NULL;
> +}
> +
> +void vsbi_handle_ecall(struct cpu_user_regs *regs)
> +{
> +    const unsigned long eid = regs->a7;
> +    const unsigned long fid = regs->a6;
> +    const struct vsbi_ext *ext = vsbi_find_extension(eid);
> +    int ret;
> +
> +    if ( ext )
> +        ret = ext->handler(eid, fid, regs);
> +    else
> +    {
> +        gprintk(XENLOG_ERR, "Unsupported Guest SBI EID #%#lx, FID #%lu\n",
> +                eid, regs->a1);

... I still consider this too verbose. Before switching RISC-V into "supported"
status, changing this and alike to gdprintk() may want considering.

Jan

Re: [PATCH v3 2/3] xen/riscv: add RISC-V legacy SBI extension support for guests

[Jan Beulich]

On 22.12.2025 17:37, Oleksii Kurochko wrote:
> This commit adds support for legacy SBI extensions (version 0.1) in Xen
> for guest domains.
> 
> The changes include:
> 1. Define all legacy SBI extension IDs (0x0 to 0x8) for better clarity and
>    completeness.
> 2. Implement handling of legacy SBI extensions, starting with support for
>    SBI_EXT_0_1_CONSOLE_PUTCHAR. SBI_EXT_0_1_CONSOLE_GETCHAR is marked as
>    not supported as legacy SBI console related stuff is expected to be used
>    only for early debugging of guest.
> 
> The implementation uses the existing virtual SBI framework to handle legacy
> SBI ecalls, ensuring compatibility with older SBI specifications in
> RISC-V guests.
> 
> Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>

Acked-by: Jan Beulich <jbeulich@suse.com>

Peeking ahead into patch 3, I notice that there you (properly) avoid saying
"this commit" in the description. Unlike here and in patch 1. Please try to
remember to avoid this and alike wording.

Jan

Re: [PATCH v3 2/3] xen/riscv: add RISC-V legacy SBI extension support for guests

[Jan Beulich]

On 22.12.2025 17:37, Oleksii Kurochko wrote:
> +static int vsbi_legacy_ecall_handler(unsigned long eid, unsigned long fid,
> +                                     struct cpu_user_regs *regs)
> +{
> +    int ret = 0;
> +
> +    switch ( eid )
> +    {
> +    case SBI_EXT_0_1_CONSOLE_PUTCHAR:
> +        vsbi_print_char(regs->a0);
> +        break;
> +
> +    case SBI_EXT_0_1_CONSOLE_GETCHAR:
> +        ret = SBI_ERR_NOT_SUPPORTED;
> +        break;
> +
> +    default:
> +        /*
> +         * TODO: domain_crash() is acceptable here while things are still under
> +         * development.
> +         * It shouldn't stay like this in the end though: guests should not
> +         * be punished like this for something Xen hasn't implemented.
> +         */
> +        domain_crash(current->domain,
> +                     "%s: Unsupported ecall: FID: #%lx, EID: #%lx\n",

Hmm, wait - patch 1 says you would consistently use #%lu for FIDs. I can adjust
while committing, unless you tell me not to.

Jan

Re: [PATCH v3 3/3] xen/riscv: add RISC-V virtual SBI base extension support for guests

[Jan Beulich]

On 22.12.2025 17:37, Oleksii Kurochko wrote:
> --- a/xen/arch/riscv/include/asm/sbi.h
> +++ b/xen/arch/riscv/include/asm/sbi.h
> @@ -14,6 +14,13 @@
>  
>  #include <xen/cpumask.h>
>  
> +/* Xen-controlled SBI version reported to guests */
> +#define XEN_SBI_VER_MAJOR 0
> +#define XEN_SBI_VER_MINOR 2

Are these going to gain a 2nd use, justifying their placement here?

> --- /dev/null
> +++ b/xen/arch/riscv/vsbi/base-extension.c
> @@ -0,0 +1,78 @@
> +
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +
> +#include <xen/lib.h>
> +#include <xen/sched.h>
> +#include <xen/version.h>
> +
> +#include <asm/processor.h>
> +#include <asm/sbi.h>
> +#include <asm/vsbi.h>
> +
> +static int vsbi_base_ecall_handler(unsigned long eid, unsigned long fid,
> +                                   struct cpu_user_regs *regs)
> +{
> +    int ret = 0;
> +    struct sbiret sbi_ret;
> +
> +    ASSERT(eid == SBI_EXT_BASE);
> +
> +    switch ( fid )
> +    {
> +    case SBI_EXT_BASE_GET_SPEC_VERSION:
> +        regs->a1 = MASK_INSR(XEN_SBI_VER_MAJOR, SBI_SPEC_VERSION_MAJOR_MASK) |
> +                   XEN_SBI_VER_MINOR;
> +        break;
> +
> +    case SBI_EXT_BASE_GET_IMP_ID:
> +        regs->a1 = SBI_XEN_IMPID;
> +        break;
> +
> +    case SBI_EXT_BASE_GET_IMP_VERSION:
> +        regs->a1 = (xen_major_version() << 16) | xen_minor_version();
> +        break;
> +
> +    case SBI_EXT_BASE_GET_MVENDORID:
> +    case SBI_EXT_BASE_GET_MARCHID:
> +    case SBI_EXT_BASE_GET_MIMPID:
> +        if ( is_hardware_domain(current->domain) )
> +        {
> +            sbi_ret = sbi_ecall(SBI_EXT_BASE, fid, 0, 0, 0, 0, 0, 0);
> +            ret = sbi_ret.error;
> +            regs->a1 = sbi_ret.value;
> +        }
> +        else
> +            /*
> +             * vSBI should present a consistent, virtualized view to guests.
> +             * In particular, DomU-visible data must remain stable across
> +             * migration and must not expose hardware-specific details.
> +             *
> +             * These register(s) must be readable in any implementation,
> +             * but a value of 0 can be returned to indicate the field
> +             * is not implemented.
> +             */
> +            regs->a1 = 0;
> +
> +        break;
> +
> +    case SBI_EXT_BASE_PROBE_EXT:
> +        regs->a1 = vsbi_find_extension(regs->a0) ? 1 : 0;
> +        break;
> +
> +    default:
> +        /*
> +         * TODO: domain_crash() is acceptable here while things are still under
> +         * development.
> +         * It shouldn't stay like this in the end though: guests should not
> +         * be punished like this for something Xen hasn't implemented.
> +         */
> +        domain_crash(current->domain,
> +                     "%s: Unsupported ecall: FID: #%lx, EID: #%lx\n",

Same remark here as for patch 2.

Jan

Re: [PATCH v3 2/3] xen/riscv: add RISC-V legacy SBI extension support for guests

[Oleksii Kurochko]

On 12/23/25 5:11 PM, Jan Beulich wrote:
> On 22.12.2025 17:37, Oleksii Kurochko wrote:
>> +static int vsbi_legacy_ecall_handler(unsigned long eid, unsigned long fid,
>> +                                     struct cpu_user_regs *regs)
>> +{
>> +    int ret = 0;
>> +
>> +    switch ( eid )
>> +    {
>> +    case SBI_EXT_0_1_CONSOLE_PUTCHAR:
>> +        vsbi_print_char(regs->a0);
>> +        break;
>> +
>> +    case SBI_EXT_0_1_CONSOLE_GETCHAR:
>> +        ret = SBI_ERR_NOT_SUPPORTED;
>> +        break;
>> +
>> +    default:
>> +        /*
>> +         * TODO: domain_crash() is acceptable here while things are still under
>> +         * development.
>> +         * It shouldn't stay like this in the end though: guests should not
>> +         * be punished like this for something Xen hasn't implemented.
>> +         */
>> +        domain_crash(current->domain,
>> +                     "%s: Unsupported ecall: FID: #%lx, EID: #%lx\n",
> Hmm, wait - patch 1 says you would consistently use #%lu for FIDs. I can adjust
> while committing, unless you tell me not to.

I think that we should drop printing FID at all for Legacy extension as according to
the spec.:
   The SBI function ID field in a6 register is ignored because these are encoded as multiple SBI
   extension IDs.
And according to "Function Listing" FID will be 0:
   https://github.com/riscv-non-isa/riscv-sbi-doc/blob/master/src/ext-legacy.adoc#function-listing

I would be happy if you could drop printing of FID during commit. Let me know if you want me to
drop printing of FID in the next patch series version.

Thanks.

~ Oleksii

Re: [PATCH v3 3/3] xen/riscv: add RISC-V virtual SBI base extension support for guests

[Oleksii Kurochko]

On 12/23/25 5:13 PM, Jan Beulich wrote:
> On 22.12.2025 17:37, Oleksii Kurochko wrote:
>> --- a/xen/arch/riscv/include/asm/sbi.h
>> +++ b/xen/arch/riscv/include/asm/sbi.h
>> @@ -14,6 +14,13 @@
>>   
>>   #include <xen/cpumask.h>
>>   
>> +/* Xen-controlled SBI version reported to guests */
>> +#define XEN_SBI_VER_MAJOR 0
>> +#define XEN_SBI_VER_MINOR 2
> Are these going to gain a 2nd use, justifying their placement here?

Good point. I don't have any plans now to use them somewhere else, so,
at least, for now it would be really put them to base-extension.c.

>
>> --- /dev/null
>> +++ b/xen/arch/riscv/vsbi/base-extension.c
>> @@ -0,0 +1,78 @@
>> +
>> +/* SPDX-License-Identifier: GPL-2.0-only */
>> +
>> +#include <xen/lib.h>
>> +#include <xen/sched.h>
>> +#include <xen/version.h>
>> +
>> +#include <asm/processor.h>
>> +#include <asm/sbi.h>
>> +#include <asm/vsbi.h>
>> +
>> +static int vsbi_base_ecall_handler(unsigned long eid, unsigned long fid,
>> +                                   struct cpu_user_regs *regs)
>> +{
>> +    int ret = 0;
>> +    struct sbiret sbi_ret;
>> +
>> +    ASSERT(eid == SBI_EXT_BASE);
>> +
>> +    switch ( fid )
>> +    {
>> +    case SBI_EXT_BASE_GET_SPEC_VERSION:
>> +        regs->a1 = MASK_INSR(XEN_SBI_VER_MAJOR, SBI_SPEC_VERSION_MAJOR_MASK) |
>> +                   XEN_SBI_VER_MINOR;
>> +        break;
>> +
>> +    case SBI_EXT_BASE_GET_IMP_ID:
>> +        regs->a1 = SBI_XEN_IMPID;
>> +        break;
>> +
>> +    case SBI_EXT_BASE_GET_IMP_VERSION:
>> +        regs->a1 = (xen_major_version() << 16) | xen_minor_version();
>> +        break;
>> +
>> +    case SBI_EXT_BASE_GET_MVENDORID:
>> +    case SBI_EXT_BASE_GET_MARCHID:
>> +    case SBI_EXT_BASE_GET_MIMPID:
>> +        if ( is_hardware_domain(current->domain) )
>> +        {
>> +            sbi_ret = sbi_ecall(SBI_EXT_BASE, fid, 0, 0, 0, 0, 0, 0);
>> +            ret = sbi_ret.error;
>> +            regs->a1 = sbi_ret.value;
>> +        }
>> +        else
>> +            /*
>> +             * vSBI should present a consistent, virtualized view to guests.
>> +             * In particular, DomU-visible data must remain stable across
>> +             * migration and must not expose hardware-specific details.
>> +             *
>> +             * These register(s) must be readable in any implementation,
>> +             * but a value of 0 can be returned to indicate the field
>> +             * is not implemented.
>> +             */
>> +            regs->a1 = 0;
>> +
>> +        break;
>> +
>> +    case SBI_EXT_BASE_PROBE_EXT:
>> +        regs->a1 = vsbi_find_extension(regs->a0) ? 1 : 0;
>> +        break;
>> +
>> +    default:
>> +        /*
>> +         * TODO: domain_crash() is acceptable here while things are still under
>> +         * development.
>> +         * It shouldn't stay like this in the end though: guests should not
>> +         * be punished like this for something Xen hasn't implemented.
>> +         */
>> +        domain_crash(current->domain,
>> +                     "%s: Unsupported ecall: FID: #%lx, EID: #%lx\n",
> Same remark here as for patch 2.

I'll update to #%lu for FID.

Thanks.

~ Oleksii

Re: [PATCH v3 2/3] xen/riscv: add RISC-V legacy SBI extension support for guests

[Oleksii Kurochko]

On 12/24/25 10:05 AM, Oleksii Kurochko wrote:
>
> On 12/23/25 5:11 PM, Jan Beulich wrote:
>> On 22.12.2025 17:37, Oleksii Kurochko wrote:
>>> +static int vsbi_legacy_ecall_handler(unsigned long eid, unsigned 
>>> long fid,
>>> +                                     struct cpu_user_regs *regs)
>>> +{
>>> +    int ret = 0;
>>> +
>>> +    switch ( eid )
>>> +    {
>>> +    case SBI_EXT_0_1_CONSOLE_PUTCHAR:
>>> +        vsbi_print_char(regs->a0);
>>> +        break;
>>> +
>>> +    case SBI_EXT_0_1_CONSOLE_GETCHAR:
>>> +        ret = SBI_ERR_NOT_SUPPORTED;
>>> +        break;
>>> +
>>> +    default:
>>> +        /*
>>> +         * TODO: domain_crash() is acceptable here while things are 
>>> still under
>>> +         * development.
>>> +         * It shouldn't stay like this in the end though: guests 
>>> should not
>>> +         * be punished like this for something Xen hasn't implemented.
>>> +         */
>>> +        domain_crash(current->domain,
>>> +                     "%s: Unsupported ecall: FID: #%lx, EID: #%lx\n",
>> Hmm, wait - patch 1 says you would consistently use #%lu for FIDs. I 
>> can adjust
>> while committing, unless you tell me not to.
>
> I think that we should drop printing FID at all for Legacy extension 
> as according to
> the spec.:
>   The SBI function ID field in a6 register is ignored because these 
> are encoded as multiple SBI
>   extension IDs.
> And according to "Function Listing" FID will be 0:
> https://github.com/riscv-non-isa/riscv-sbi-doc/blob/master/src/ext-legacy.adoc#function-listing
>
> I would be happy if you could drop printing of FID during commit. Let 
> me know if you want me to
> drop printing of FID in the next patch series version.
>
EID printing should be also updated to #%#lx.

Sorry for noticing that only now.

~ Oleksii