* [PULL 1/4] tests/tcg: Replace -mpower8-vector with -mcpu=power8
2024-11-05 15:03 [PULL 0/4] linux-user patch queue Richard Henderson
@ 2024-11-05 15:03 ` Richard Henderson
2024-11-05 15:03 ` [PULL 2/4] linux-user: Fix GDB complaining about system-supplied DSO string table index Richard Henderson
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Richard Henderson @ 2024-11-05 15:03 UTC (permalink / raw)
To: qemu-devel; +Cc: Ilya Leoshkevich, Alex Bennée
From: Ilya Leoshkevich <iii@linux.ibm.com>
[1] deprecated -mpower8-vector, resulting in:
powerpc64-linux-gnu-gcc: warning: switch '-mpower8-vector' is no longer supported
qemu/tests/tcg/ppc64/vsx_f2i_nan.c:4:15: error: expected ';' before 'float'
4 | typedef vector float vsx_float32_vec_t;
| ^~~~~~
Use -mcpu=power8 instead. In order to properly verify that this works,
one needs a big-endian (the minimum supported CPU for 64-bit
little-endian is power8 anyway) GCC configured with --enable-checking
(see GCC commit e154242724b0 ("[RS6000] Don't pass -many to the
assembler").
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109987
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Acked-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20241023131250.48510-1-iii@linux.ibm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tests/tcg/ppc64/Makefile.target | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/tests/tcg/ppc64/Makefile.target b/tests/tcg/ppc64/Makefile.target
index 1940886c73..0d058b2600 100644
--- a/tests/tcg/ppc64/Makefile.target
+++ b/tests/tcg/ppc64/Makefile.target
@@ -6,7 +6,7 @@ VPATH += $(SRC_PATH)/tests/tcg/ppc64
config-cc.mak: Makefile
$(quiet-@)( \
- $(call cc-option,-mpower8-vector, CROSS_CC_HAS_POWER8_VECTOR); \
+ $(call cc-option,-mcpu=power8, CROSS_CC_HAS_CPU_POWER8); \
$(call cc-option,-mpower10, CROSS_CC_HAS_POWER10)) 3> config-cc.mak
-include config-cc.mak
@@ -23,15 +23,15 @@ run-threadcount: threadcount
run-plugin-threadcount-with-%:
$(call skip-test, $<, "BROKEN (flaky with clang) ")
-ifneq ($(CROSS_CC_HAS_POWER8_VECTOR),)
+ifneq ($(CROSS_CC_HAS_CPU_POWER8),)
PPC64_TESTS=bcdsub non_signalling_xscv
endif
-$(PPC64_TESTS): CFLAGS += -mpower8-vector
+$(PPC64_TESTS): CFLAGS += -mcpu=power8
-ifneq ($(CROSS_CC_HAS_POWER8_VECTOR),)
+ifneq ($(CROSS_CC_HAS_CPU_POWER8),)
PPC64_TESTS += vsx_f2i_nan
endif
-vsx_f2i_nan: CFLAGS += -mpower8-vector -I$(SRC_PATH)/include
+vsx_f2i_nan: CFLAGS += -mcpu=power8 -I$(SRC_PATH)/include
PPC64_TESTS += mtfsf
PPC64_TESTS += mffsce
--
2.43.0
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PULL 2/4] linux-user: Fix GDB complaining about system-supplied DSO string table index
2024-11-05 15:03 [PULL 0/4] linux-user patch queue Richard Henderson
2024-11-05 15:03 ` [PULL 1/4] tests/tcg: Replace -mpower8-vector with -mcpu=power8 Richard Henderson
@ 2024-11-05 15:03 ` Richard Henderson
2024-11-05 15:03 ` [PULL 3/4] linux-user: Allow custom rt signal mappings Richard Henderson
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Richard Henderson @ 2024-11-05 15:03 UTC (permalink / raw)
To: qemu-devel; +Cc: Ilya Leoshkevich
From: Ilya Leoshkevich <iii@linux.ibm.com>
When debugging qemu-user processes using gdbstub, the following warning
appears every time:
warning: BFD: warning: system-supplied DSO at 0x7f8253cc3000 has a corrupt string table index
The reason is that QEMU does not map the VDSO's section headers. The
VDSO's ELF header's e_shoff points to zeros, which GDB fails to parse.
The difference with the kernel's VDSO is that the latter is mapped as a
blob, ignoring program headers - which also don't cover the section
table. QEMU, on the other hand, loads it as an ELF file.
There appears to be no way to place section headers inside a section,
and, therefore, no way to refer to them from a linker script. Also, ld
hardcodes section headers to be non-loadable, see
_bfd_elf_assign_file_positions_for_non_load(). In theory ld could be
enhanced by implementing an "SHDRS" keyword in addition to the existing
"FILEHDR" and "PHDRS".
There are multiple ways to resolve the issue:
- Copy VDSO as a blob in load_elf_vdso(). This would require creating
specialized loader logic, that duplicates parts of load_elf_image().
- Fix up VDSO's PHDR size in load_elf_vdso(). This would require either
duplicating the parsing logic, or adding an ugly parameter to
load_elf_image().
- Fix up VDSO's PHDR size in gen-vdso. This is the simplest solution,
so do it.
There are two tricky parts:
- Byte-swaps need to be done either on local copies, or in-place and
then reverted in the end. To preserve the existing code structure, do
the former for Sym and Dyn, and the latter for Ehdr, Phdr, and Shdr.
- There must be no .bss, which is already the case - but having an
explicit check is helpful to ensure correctness.
To verify this change, I diffed the on-disk and the loaded VDSOs; the
result does not show anything unusual, except for what seems to be an
existing oversight (which should probably be fixed separately):
│ Symbol table '.dynsym' contains 8 entries:
│ Num: Value Size Type Bind Vis Ndx Name
│ - 0: 0000000000000000 0 NOTYPE LOCAL DEFAULT UND
│ - 6: 0000000000000000 0 OBJECT GLOBAL DEFAULT ABS LINUX_2.6.29
│ + 0: 00007f61075bf000 0 NOTYPE LOCAL DEFAULT UND
│ + 6: 00007f61075bf000 0 OBJECT GLOBAL DEFAULT ABS LINUX_2.6.29
Fixes: 2fa536d10797 ("linux-user: Add gen-vdso tool")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20241023202850.55211-1-iii@linux.ibm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
linux-user/gen-vdso.c | 40 +++++++-------
linux-user/gen-vdso-elfn.c.inc | 98 +++++++++++++++++++++++-----------
2 files changed, 87 insertions(+), 51 deletions(-)
diff --git a/linux-user/gen-vdso.c b/linux-user/gen-vdso.c
index 31e333be80..721f38d5a3 100644
--- a/linux-user/gen-vdso.c
+++ b/linux-user/gen-vdso.c
@@ -131,23 +131,6 @@ int main(int argc, char **argv)
}
fclose(inf);
- /*
- * Write out the vdso image now, before we make local changes.
- */
-
- fprintf(outf,
- "/* Automatically generated from linux-user/gen-vdso.c. */\n"
- "\n"
- "static const uint8_t %s_image[] = {",
- prefix);
- for (long i = 0; i < total_len; ++i) {
- if (i % 12 == 0) {
- fputs("\n ", outf);
- }
- fprintf(outf, " 0x%02x,", buf[i]);
- }
- fprintf(outf, "\n};\n\n");
-
/*
* Identify which elf flavor we're processing.
* The first 16 bytes of the file are e_ident.
@@ -179,14 +162,17 @@ int main(int argc, char **argv)
* Output relocation addresses as we go.
*/
- fprintf(outf, "static const unsigned %s_relocs[] = {\n", prefix);
+ fprintf(outf,
+ "/* Automatically generated by linux-user/gen-vdso.c. */\n"
+ "\n"
+ "static const unsigned %s_relocs[] = {\n", prefix);
switch (buf[EI_CLASS]) {
case ELFCLASS32:
- elf32_process(outf, buf, need_bswap);
+ elf32_process(outf, buf, total_len, need_bswap);
break;
case ELFCLASS64:
- elf64_process(outf, buf, need_bswap);
+ elf64_process(outf, buf, total_len, need_bswap);
break;
default:
fprintf(stderr, "%s: invalid elf EI_CLASS (%u)\n",
@@ -196,6 +182,20 @@ int main(int argc, char **argv)
fprintf(outf, "};\n\n"); /* end vdso_relocs. */
+ /*
+ * Write out the vdso image now, after we made local changes.
+ */
+ fprintf(outf,
+ "static const uint8_t %s_image[] = {",
+ prefix);
+ for (long i = 0; i < total_len; ++i) {
+ if (i % 12 == 0) {
+ fputs("\n ", outf);
+ }
+ fprintf(outf, " 0x%02x,", buf[i]);
+ }
+ fprintf(outf, "\n};\n\n");
+
fprintf(outf, "static const VdsoImageInfo %s_image_info = {\n", prefix);
fprintf(outf, " .image = %s_image,\n", prefix);
fprintf(outf, " .relocs = %s_relocs,\n", prefix);
diff --git a/linux-user/gen-vdso-elfn.c.inc b/linux-user/gen-vdso-elfn.c.inc
index 95856eb839..b47019e136 100644
--- a/linux-user/gen-vdso-elfn.c.inc
+++ b/linux-user/gen-vdso-elfn.c.inc
@@ -68,28 +68,45 @@ static void elfN(search_symtab)(ElfN(Shdr) *shdr, unsigned sym_idx,
void *buf, bool need_bswap)
{
unsigned str_idx = shdr[sym_idx].sh_link;
- ElfN(Sym) *sym = buf + shdr[sym_idx].sh_offset;
- unsigned sym_n = shdr[sym_idx].sh_size / sizeof(*sym);
+ ElfN(Sym) *target_sym = buf + shdr[sym_idx].sh_offset;
+ unsigned sym_n = shdr[sym_idx].sh_size / sizeof(*target_sym);
const char *str = buf + shdr[str_idx].sh_offset;
for (unsigned i = 0; i < sym_n; ++i) {
const char *name;
+ ElfN(Sym) sym;
+ memcpy(&sym, &target_sym[i], sizeof(sym));
if (need_bswap) {
- elfN(bswap_sym)(sym + i);
+ elfN(bswap_sym)(&sym);
}
- name = str + sym[i].st_name;
+ name = str + sym.st_name;
if (sigreturn_sym && strcmp(sigreturn_sym, name) == 0) {
- sigreturn_addr = sym[i].st_value;
+ sigreturn_addr = sym.st_value;
}
if (rt_sigreturn_sym && strcmp(rt_sigreturn_sym, name) == 0) {
- rt_sigreturn_addr = sym[i].st_value;
+ rt_sigreturn_addr = sym.st_value;
}
}
}
-static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
+static void elfN(bswap_ps_hdrs)(ElfN(Ehdr) *ehdr)
+{
+ ElfN(Phdr) *phdr = (void *)ehdr + ehdr->e_phoff;
+ ElfN(Shdr) *shdr = (void *)ehdr + ehdr->e_shoff;
+ ElfN(Half) i;
+
+ for (i = 0; i < ehdr->e_phnum; ++i) {
+ elfN(bswap_phdr)(&phdr[i]);
+ }
+
+ for (i = 0; i < ehdr->e_shnum; ++i) {
+ elfN(bswap_shdr)(&shdr[i]);
+ }
+}
+
+static void elfN(process)(FILE *outf, void *buf, long len, bool need_bswap)
{
ElfN(Ehdr) *ehdr = buf;
ElfN(Phdr) *phdr;
@@ -103,24 +120,14 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
int errors = 0;
if (need_bswap) {
- elfN(bswap_ehdr)(ehdr);
+ elfN(bswap_ehdr)(buf);
+ elfN(bswap_ps_hdrs)(buf);
}
phnum = ehdr->e_phnum;
phdr = buf + ehdr->e_phoff;
- if (need_bswap) {
- for (unsigned i = 0; i < phnum; ++i) {
- elfN(bswap_phdr)(phdr + i);
- }
- }
-
shnum = ehdr->e_shnum;
shdr = buf + ehdr->e_shoff;
- if (need_bswap) {
- for (unsigned i = 0; i < shnum; ++i) {
- elfN(bswap_shdr)(shdr + i);
- }
- }
for (unsigned i = 0; i < shnum; ++i) {
switch (shdr[i].sh_type) {
case SHT_SYMTAB:
@@ -154,7 +161,24 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
fprintf(stderr, "LOAD segment not loaded at address 0\n");
errors++;
}
- first_segsz = phdr[i].p_filesz;
+ /*
+ * Extend the program header to cover the entire VDSO, so that
+ * load_elf_vdso() loads everything, including section headers.
+ *
+ * Require that there is no .bss, since it would break this
+ * approach.
+ */
+ if (phdr[i].p_filesz != phdr[i].p_memsz) {
+ fprintf(stderr, "LOAD segment's filesz and memsz differ\n");
+ errors++;
+ }
+ if (phdr[i].p_filesz > len) {
+ fprintf(stderr, "LOAD segment is larger than the whole VDSO\n");
+ errors++;
+ }
+ phdr[i].p_filesz = len;
+ phdr[i].p_memsz = len;
+ first_segsz = len;
if (first_segsz < ehdr->e_phoff + phnum * sizeof(*phdr)) {
fprintf(stderr, "LOAD segment does not cover PHDRs\n");
errors++;
@@ -197,17 +221,24 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
output_reloc(outf, buf, &phdr[i].p_paddr);
}
+ /* Relocate the section headers. */
+ for (unsigned i = 0; i < shnum; ++i) {
+ output_reloc(outf, buf, &shdr[i].sh_addr);
+ }
+
/* Relocate the DYNAMIC entries. */
if (dynamic_addr) {
- ElfN(Dyn) *dyn = buf + dynamic_ofs;
- __typeof(dyn->d_tag) tag;
+ ElfN(Dyn) *target_dyn = buf + dynamic_ofs;
+ __typeof(((ElfN(Dyn) *)target_dyn)->d_tag) tag;
do {
+ ElfN(Dyn) dyn;
+ memcpy(&dyn, target_dyn, sizeof(dyn));
if (need_bswap) {
- elfN(bswap_dyn)(dyn);
+ elfN(bswap_dyn)(&dyn);
}
- tag = dyn->d_tag;
+ tag = dyn.d_tag;
switch (tag) {
case DT_HASH:
@@ -218,7 +249,7 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
case DT_PLTGOT:
case DT_ADDRRNGLO ... DT_ADDRRNGHI:
/* These entries store an address in the entry. */
- output_reloc(outf, buf, &dyn->d_un.d_val);
+ output_reloc(outf, buf, &target_dyn->d_un.d_val);
break;
case DT_NULL:
@@ -235,7 +266,7 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
break;
case DT_SYMENT:
- if (dyn->d_un.d_val != sizeof(ElfN(Sym))) {
+ if (dyn.d_un.d_val != sizeof(ElfN(Sym))) {
fprintf(stderr, "VDSO has incorrect dynamic symbol size\n");
errors++;
}
@@ -251,7 +282,7 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
* ??? The RISC-V toolchain will emit these even when there
* are no relocations. Validate zeros.
*/
- if (dyn->d_un.d_val != 0) {
+ if (dyn.d_un.d_val != 0) {
fprintf(stderr, "VDSO has dynamic relocations\n");
errors++;
}
@@ -287,7 +318,7 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
errors++;
break;
}
- dyn++;
+ target_dyn++;
} while (tag != DT_NULL);
if (errors) {
exit(EXIT_FAILURE);
@@ -296,11 +327,11 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
/* Relocate the dynamic symbol table. */
if (dynsym_idx) {
- ElfN(Sym) *sym = buf + shdr[dynsym_idx].sh_offset;
- unsigned sym_n = shdr[dynsym_idx].sh_size / sizeof(*sym);
+ ElfN(Sym) *target_sym = buf + shdr[dynsym_idx].sh_offset;
+ unsigned sym_n = shdr[dynsym_idx].sh_size / sizeof(*target_sym);
for (unsigned i = 0; i < sym_n; ++i) {
- output_reloc(outf, buf, &sym[i].st_value);
+ output_reloc(outf, buf, &target_sym[i].st_value);
}
}
@@ -311,4 +342,9 @@ static void elfN(process)(FILE *outf, void *buf, bool need_bswap)
if (symtab_idx) {
elfN(search_symtab)(shdr, symtab_idx, buf, need_bswap);
}
+
+ if (need_bswap) {
+ elfN(bswap_ps_hdrs)(buf);
+ elfN(bswap_ehdr)(buf);
+ }
}
--
2.43.0
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PULL 3/4] linux-user: Allow custom rt signal mappings
2024-11-05 15:03 [PULL 0/4] linux-user patch queue Richard Henderson
2024-11-05 15:03 ` [PULL 1/4] tests/tcg: Replace -mpower8-vector with -mcpu=power8 Richard Henderson
2024-11-05 15:03 ` [PULL 2/4] linux-user: Fix GDB complaining about system-supplied DSO string table index Richard Henderson
@ 2024-11-05 15:03 ` Richard Henderson
2024-11-05 15:03 ` [PULL 4/4] tests/tcg: Add SIGRTMIN/SIGRTMAX test Richard Henderson
2024-11-06 17:27 ` [PULL 0/4] linux-user patch queue Peter Maydell
4 siblings, 0 replies; 6+ messages in thread
From: Richard Henderson @ 2024-11-05 15:03 UTC (permalink / raw)
To: qemu-devel; +Cc: Ilya Leoshkevich
From: Ilya Leoshkevich <iii@linux.ibm.com>
Some applications want to use low priority realtime signals (e.g.,
SIGRTMAX). Currently QEMU cannot map all target realtime signals to
host realtime signals, and chooses to sacrifice the end of the target
realtime signal range.
Allow users to choose how to map target realtime signals to host
realtime signals using the new -t option, the new QEMU_RTSIG_MAP
environment variable, and the new -Drtsig_map=\"...\" meson flag.
To simplify things, the meson flag is not per-target, because the
intended use case is app-specific qemu-user builds.
The mapping is specified using the "tsig hsig count[,...]" syntax.
Target realtime signals [tsig,tsig+count) are mapped to host realtime
signals [hsig,hsig+count). Care is taken to avoid double and
out-of-range mappings.
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20241029232211.206766-2-iii@linux.ibm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
linux-user/signal-common.h | 2 +-
linux-user/main.c | 12 +++-
linux-user/signal.c | 108 +++++++++++++++++++++++++++-------
meson.build | 3 +-
meson_options.txt | 2 +
scripts/meson-buildoptions.sh | 2 +
6 files changed, 106 insertions(+), 23 deletions(-)
diff --git a/linux-user/signal-common.h b/linux-user/signal-common.h
index f4cbe6185e..8584d9ecc2 100644
--- a/linux-user/signal-common.h
+++ b/linux-user/signal-common.h
@@ -56,7 +56,7 @@ void setup_rt_frame(int sig, struct target_sigaction *ka,
target_sigset_t *set, CPUArchState *env);
void process_pending_signals(CPUArchState *cpu_env);
-void signal_init(void);
+void signal_init(const char *rtsig_map);
void queue_signal(CPUArchState *env, int sig, int si_type,
target_siginfo_t *info);
void host_to_target_siginfo(target_siginfo_t *tinfo, const siginfo_t *info);
diff --git a/linux-user/main.c b/linux-user/main.c
index 8143a0d4b0..b09af8d436 100644
--- a/linux-user/main.c
+++ b/linux-user/main.c
@@ -412,6 +412,13 @@ static void handle_arg_reserved_va(const char *arg)
reserved_va = val ? val - 1 : 0;
}
+static const char *rtsig_map = CONFIG_QEMU_RTSIG_MAP;
+
+static void handle_arg_rtsig_map(const char *arg)
+{
+ rtsig_map = arg;
+}
+
static void handle_arg_one_insn_per_tb(const char *arg)
{
opt_one_insn_per_tb = true;
@@ -494,6 +501,9 @@ static const struct qemu_argument arg_table[] = {
"address", "set guest_base address to 'address'"},
{"R", "QEMU_RESERVED_VA", true, handle_arg_reserved_va,
"size", "reserve 'size' bytes for guest virtual address space"},
+ {"t", "QEMU_RTSIG_MAP", true, handle_arg_rtsig_map,
+ "tsig hsig n[,...]",
+ "map target rt signals [tsig,tsig+n) to [hsig,hsig+n]"},
{"d", "QEMU_LOG", true, handle_arg_log,
"item[,...]", "enable logging of specified items "
"(use '-d help' for a list of items)"},
@@ -1002,7 +1012,7 @@ int main(int argc, char **argv, char **envp)
target_set_brk(info->brk);
syscall_init();
- signal_init();
+ signal_init(rtsig_map);
/* Now that we've loaded the binary, GUEST_BASE is fixed. Delay
generating the prologue until now so that the prologue can take
diff --git a/linux-user/signal.c b/linux-user/signal.c
index 63ac2df53b..9b6d772882 100644
--- a/linux-user/signal.c
+++ b/linux-user/signal.c
@@ -18,6 +18,7 @@
*/
#include "qemu/osdep.h"
#include "qemu/bitops.h"
+#include "qemu/cutils.h"
#include "gdbstub/user.h"
#include "exec/page-protection.h"
#include "hw/core/tcg-cpu-ops.h"
@@ -513,20 +514,81 @@ static int core_dump_signal(int sig)
}
}
-static void signal_table_init(void)
+static void signal_table_init(const char *rtsig_map)
{
int hsig, tsig, count;
+ if (rtsig_map) {
+ /*
+ * Map host RT signals to target RT signals according to the
+ * user-provided specification.
+ */
+ const char *s = rtsig_map;
+
+ while (true) {
+ int i;
+
+ if (qemu_strtoi(s, &s, 10, &tsig) || *s++ != ' ') {
+ fprintf(stderr, "Malformed target signal in QEMU_RTSIG_MAP\n");
+ exit(EXIT_FAILURE);
+ }
+ if (qemu_strtoi(s, &s, 10, &hsig) || *s++ != ' ') {
+ fprintf(stderr, "Malformed host signal in QEMU_RTSIG_MAP\n");
+ exit(EXIT_FAILURE);
+ }
+ if (qemu_strtoi(s, &s, 10, &count) || (*s && *s != ',')) {
+ fprintf(stderr, "Malformed signal count in QEMU_RTSIG_MAP\n");
+ exit(EXIT_FAILURE);
+ }
+
+ for (i = 0; i < count; i++, tsig++, hsig++) {
+ if (tsig < TARGET_SIGRTMIN || tsig > TARGET_NSIG) {
+ fprintf(stderr, "%d is not a target rt signal\n", tsig);
+ exit(EXIT_FAILURE);
+ }
+ if (hsig < SIGRTMIN || hsig > SIGRTMAX) {
+ fprintf(stderr, "%d is not a host rt signal\n", hsig);
+ exit(EXIT_FAILURE);
+ }
+ if (host_to_target_signal_table[hsig]) {
+ fprintf(stderr, "%d already maps %d\n",
+ hsig, host_to_target_signal_table[hsig]);
+ exit(EXIT_FAILURE);
+ }
+ host_to_target_signal_table[hsig] = tsig;
+ }
+
+ if (*s) {
+ s++;
+ } else {
+ break;
+ }
+ }
+ } else {
+ /*
+ * Default host-to-target RT signal mapping.
+ *
+ * Signals are supported starting from TARGET_SIGRTMIN and going up
+ * until we run out of host realtime signals. Glibc uses the lower 2
+ * RT signals and (hopefully) nobody uses the upper ones.
+ * This is why SIGRTMIN (34) is generally greater than __SIGRTMIN (32).
+ * To fix this properly we would need to do manual signal delivery
+ * multiplexed over a single host signal.
+ * Attempts for configure "missing" signals via sigaction will be
+ * silently ignored.
+ *
+ * Reserve one signal for internal usage (see below).
+ */
+
+ hsig = SIGRTMIN + 1;
+ for (tsig = TARGET_SIGRTMIN;
+ hsig <= SIGRTMAX && tsig <= TARGET_NSIG;
+ hsig++, tsig++) {
+ host_to_target_signal_table[hsig] = tsig;
+ }
+ }
+
/*
- * Signals are supported starting from TARGET_SIGRTMIN and going up
- * until we run out of host realtime signals. Glibc uses the lower 2
- * RT signals and (hopefully) nobody uses the upper ones.
- * This is why SIGRTMIN (34) is generally greater than __SIGRTMIN (32).
- * To fix this properly we would need to do manual signal delivery
- * multiplexed over a single host signal.
- * Attempts for configure "missing" signals via sigaction will be
- * silently ignored.
- *
* Remap the target SIGABRT, so that we can distinguish host abort
* from guest abort. When the guest registers a signal handler or
* calls raise(SIGABRT), the host will raise SIG_RTn. If the guest
@@ -536,21 +598,27 @@ static void signal_table_init(void)
* parent sees the correct mapping from wait status.
*/
- hsig = SIGRTMIN;
host_to_target_signal_table[SIGABRT] = 0;
- host_to_target_signal_table[hsig++] = TARGET_SIGABRT;
-
- for (tsig = TARGET_SIGRTMIN;
- hsig <= SIGRTMAX && tsig <= TARGET_NSIG;
- hsig++, tsig++) {
- host_to_target_signal_table[hsig] = tsig;
+ for (hsig = SIGRTMIN; hsig <= SIGRTMAX; hsig++) {
+ if (!host_to_target_signal_table[hsig]) {
+ host_to_target_signal_table[hsig] = TARGET_SIGABRT;
+ break;
+ }
+ }
+ if (hsig > SIGRTMAX) {
+ fprintf(stderr, "No rt signals left for SIGABRT mapping\n");
+ exit(EXIT_FAILURE);
}
/* Invert the mapping that has already been assigned. */
for (hsig = 1; hsig < _NSIG; hsig++) {
tsig = host_to_target_signal_table[hsig];
if (tsig) {
- assert(target_to_host_signal_table[tsig] == 0);
+ if (target_to_host_signal_table[tsig]) {
+ fprintf(stderr, "%d is already mapped to %d\n",
+ tsig, target_to_host_signal_table[tsig]);
+ exit(EXIT_FAILURE);
+ }
target_to_host_signal_table[tsig] = hsig;
}
}
@@ -573,13 +641,13 @@ static void signal_table_init(void)
trace_signal_table_init(count);
}
-void signal_init(void)
+void signal_init(const char *rtsig_map)
{
TaskState *ts = get_task_state(thread_cpu);
struct sigaction act, oact;
/* initialize signal conversion tables */
- signal_table_init();
+ signal_table_init(rtsig_map);
/* Set the signal mask from the host mask. */
sigprocmask(0, 0, &ts->signal_mask);
diff --git a/meson.build b/meson.build
index c386593c52..9fa9d059ff 100644
--- a/meson.build
+++ b/meson.build
@@ -3178,7 +3178,8 @@ foreach target : target_dirs
config_target += {
'CONFIG_USER_ONLY': 'y',
'CONFIG_QEMU_INTERP_PREFIX':
- get_option('interp_prefix').replace('%M', config_target['TARGET_NAME'])
+ get_option('interp_prefix').replace('%M', config_target['TARGET_NAME']),
+ 'CONFIG_QEMU_RTSIG_MAP': get_option('rtsig_map'),
}
endif
diff --git a/meson_options.txt b/meson_options.txt
index 24bf009056..ac4887a622 100644
--- a/meson_options.txt
+++ b/meson_options.txt
@@ -27,6 +27,8 @@ option('block_drv_ro_whitelist', type : 'string', value : '',
description: 'set block driver read-only whitelist (by default affects only QEMU, not tools like qemu-img)')
option('interp_prefix', type : 'string', value : '/usr/gnemul/qemu-%M',
description: 'where to find shared libraries etc., use %M for cpu name')
+option('rtsig_map', type : 'string', value : 'NULL',
+ description: 'default value of QEMU_RTSIG_MAP')
option('fuzzing_engine', type : 'string', value : '',
description: 'fuzzing engine library for OSS-Fuzz')
option('trace_file', type: 'string', value: 'trace',
diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
index 6f2bb08ecd..51ed46e46a 100644
--- a/scripts/meson-buildoptions.sh
+++ b/scripts/meson-buildoptions.sh
@@ -72,6 +72,7 @@ meson_options_help() {
printf "%s\n" ' "manufacturer" name for qemu-ga registry entries'
printf "%s\n" ' [QEMU]'
printf "%s\n" ' --qemu-ga-version=VALUE version number for qemu-ga installer'
+ printf "%s\n" ' --rtsig-map=VALUE default value of QEMU_RTSIG_MAP [NULL]'
printf "%s\n" ' --smbd=VALUE Path to smbd for slirp networking'
printf "%s\n" ' --sysconfdir=VALUE Sysconf data directory [etc]'
printf "%s\n" ' --tls-priority=VALUE Default TLS protocol/cipher priority string'
@@ -460,6 +461,7 @@ _meson_option_parse() {
--disable-replication) printf "%s" -Dreplication=disabled ;;
--enable-rng-none) printf "%s" -Drng_none=true ;;
--disable-rng-none) printf "%s" -Drng_none=false ;;
+ --rtsig-map=*) quote_sh "-Drtsig_map=$2" ;;
--enable-rust) printf "%s" -Drust=enabled ;;
--disable-rust) printf "%s" -Drust=disabled ;;
--enable-rutabaga-gfx) printf "%s" -Drutabaga_gfx=enabled ;;
--
2.43.0
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PULL 4/4] tests/tcg: Add SIGRTMIN/SIGRTMAX test
2024-11-05 15:03 [PULL 0/4] linux-user patch queue Richard Henderson
` (2 preceding siblings ...)
2024-11-05 15:03 ` [PULL 3/4] linux-user: Allow custom rt signal mappings Richard Henderson
@ 2024-11-05 15:03 ` Richard Henderson
2024-11-06 17:27 ` [PULL 0/4] linux-user patch queue Peter Maydell
4 siblings, 0 replies; 6+ messages in thread
From: Richard Henderson @ 2024-11-05 15:03 UTC (permalink / raw)
To: qemu-devel; +Cc: Ilya Leoshkevich
From: Ilya Leoshkevich <iii@linux.ibm.com>
Test the lowest and the highest real-time signals. This requires
configuring the real-time signal mapping, and therefore some knowledge
about the host. To this end, pass the emulator path in the QEMU
environment variable to all tests (this should not disturb the existing
ones), and assume that all hosts have signals 36-39 available.
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-ID: <20241029232211.206766-3-iii@linux.ibm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tests/tcg/multiarch/linux/linux-sigrtminmax.c | 74 +++++++++++++++++++
tests/tcg/Makefile.target | 4 +-
2 files changed, 76 insertions(+), 2 deletions(-)
create mode 100644 tests/tcg/multiarch/linux/linux-sigrtminmax.c
diff --git a/tests/tcg/multiarch/linux/linux-sigrtminmax.c b/tests/tcg/multiarch/linux/linux-sigrtminmax.c
new file mode 100644
index 0000000000..a7059aacd9
--- /dev/null
+++ b/tests/tcg/multiarch/linux/linux-sigrtminmax.c
@@ -0,0 +1,74 @@
+/*
+ * Test the lowest and the highest real-time signals.
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+#include <assert.h>
+#include <signal.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+/* For hexagon and microblaze. */
+#ifndef __SIGRTMIN
+#define __SIGRTMIN 32
+#endif
+
+extern char **environ;
+
+static bool seen_sigrtmin, seen_sigrtmax;
+
+static void handle_signal(int sig)
+{
+ if (sig == SIGRTMIN) {
+ seen_sigrtmin = true;
+ } else if (sig == SIGRTMAX) {
+ seen_sigrtmax = true;
+ } else {
+ _exit(1);
+ }
+}
+
+int main(int argc, char **argv)
+{
+ char *qemu = getenv("QEMU");
+ struct sigaction act;
+
+ assert(qemu);
+
+ if (!getenv("QEMU_RTSIG_MAP")) {
+ char **new_argv = malloc((argc + 2) + sizeof(char *));
+ int tsig1, hsig1, count1, tsig2, hsig2, count2;
+ char rt_sigmap[64];
+
+ /* Re-exec with a mapping that includes SIGRTMIN and SIGRTMAX. */
+ new_argv[0] = qemu;
+ memcpy(&new_argv[1], argv, (argc + 1) * sizeof(char *));
+ tsig1 = __SIGRTMIN;
+ /* The host must have a few signals starting from this one. */
+ hsig1 = 36;
+ count1 = SIGRTMIN - __SIGRTMIN + 1;
+ tsig2 = SIGRTMAX;
+ hsig2 = hsig1 + count1;
+ count2 = 1;
+ snprintf(rt_sigmap, sizeof(rt_sigmap), "%d %d %d,%d %d %d",
+ tsig1, hsig1, count1, tsig2, hsig2, count2);
+ setenv("QEMU_RTSIG_MAP", rt_sigmap, 0);
+ assert(execve(new_argv[0], new_argv, environ) == 0);
+ return EXIT_FAILURE;
+ }
+
+ memset(&act, 0, sizeof(act));
+ act.sa_handler = handle_signal;
+ assert(sigaction(SIGRTMIN, &act, NULL) == 0);
+ assert(sigaction(SIGRTMAX, &act, NULL) == 0);
+
+ assert(kill(getpid(), SIGRTMIN) == 0);
+ assert(seen_sigrtmin);
+ assert(kill(getpid(), SIGRTMAX) == 0);
+ assert(seen_sigrtmax);
+
+ return EXIT_SUCCESS;
+}
diff --git a/tests/tcg/Makefile.target b/tests/tcg/Makefile.target
index 9722145b97..95ff76ea44 100644
--- a/tests/tcg/Makefile.target
+++ b/tests/tcg/Makefile.target
@@ -179,10 +179,10 @@ run-plugin-%-with-libmem.so: PLUGIN_ARGS=$(COMMA)inline=true
ifeq ($(filter %-softmmu, $(TARGET)),)
run-%: %
- $(call run-test, $<, $(QEMU) $(QEMU_OPTS) $<)
+ $(call run-test, $<, env QEMU=$(QEMU) $(QEMU) $(QEMU_OPTS) $<)
run-plugin-%:
- $(call run-test, $@, $(QEMU) $(QEMU_OPTS) \
+ $(call run-test, $@, env QEMU=$(QEMU) $(QEMU) $(QEMU_OPTS) \
-plugin $(PLUGIN_LIB)/$(call extract-plugin,$@)$(PLUGIN_ARGS) \
-d plugin -D $*.pout \
$(call strip-plugin,$<))
--
2.43.0
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PULL 0/4] linux-user patch queue
2024-11-05 15:03 [PULL 0/4] linux-user patch queue Richard Henderson
` (3 preceding siblings ...)
2024-11-05 15:03 ` [PULL 4/4] tests/tcg: Add SIGRTMIN/SIGRTMAX test Richard Henderson
@ 2024-11-06 17:27 ` Peter Maydell
4 siblings, 0 replies; 6+ messages in thread
From: Peter Maydell @ 2024-11-06 17:27 UTC (permalink / raw)
To: Richard Henderson; +Cc: qemu-devel
On Tue, 5 Nov 2024 at 15:05, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> The following changes since commit 11b8920ed2093848f79f93d106afe8a69a61a523:
>
> Merge tag 'pull-request-2024-11-04' of https://gitlab.com/thuth/qemu into staging (2024-11-04 17:37:59 +0000)
>
> are available in the Git repository at:
>
> https://gitlab.com/rth7680/qemu.git tags/pull-lu-20241105
>
> for you to fetch changes up to 6f58b090328f5b47ee688c77542bf20d521239a2:
>
> tests/tcg: Add SIGRTMIN/SIGRTMAX test (2024-11-05 10:36:08 +0000)
>
> ----------------------------------------------------------------
> tests/tcg: Replace -mpower8-vector with -mcpu=power8
> linux-user: Fix GDB complaining about system-supplied DSO string table index
> linux-user: Allow custom rt signal mappings
>
> ----------------------------------------------------------------
Applied, thanks.
Please update the changelog at https://wiki.qemu.org/ChangeLog/9.2
for any user-visible changes.
-- PMM
^ permalink raw reply [flat|nested] 6+ messages in thread