4:15 a.m.
This series fixes some build warnings seen when building gcore for analysis of
32-bit ARM dumps on a 64-bit x86 host. Without the fixes, gcore.so sometimes
segfaults due to stack corruption due to the type mismatches.
Thank you.
Vincent Whitchurch (2):
gcore: fix phys addr type
gcore: fix offset types
src/libgcore/gcore_coredump.c | 3 ++-
src/libgcore/gcore_defs.h | 2 +-
src/libgcore/gcore_elf_struct.c | 4 ++--
3 files changed, 5 insertions(+), 4 deletions(-)
--
2.28.0
4:15 a.m.
New subject: [PATCH crash-gcore 1/2] gcore: fix phys addr type
Fix these warnings with -m32 (e.g. ARM) builds on 64-bit. Without the
fix, gcore.so sometimes segfaults due to overwriting of stack variables.
libgcore/gcore_coredump.c: In function ‘gcore_readmem_user’:
libgcore/gcore_coredump.c:85:26: warning: passing argument 2 of
‘uvtop_quiet’ from incompatible pointer type
[-Wincompatible-pointer-types]
if (!uvtop_quiet(addr, &paddr)) {
^~~~~~
libgcore/gcore_coredump.c:71:49: note: expected ‘physaddr_t *’ {aka
‘long long unsigned int *’} but argument is of type ‘ulong *’ {aka ‘long
unsigned int *’}
static int uvtop_quiet(ulong vaddr, physaddr_t *paddr);
~~~~~~~~~~~~^~~~~
Signed-off-by: Vincent Whitchurch <vincent.whitchurch(a)axis.com>
---
src/libgcore/gcore_coredump.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/libgcore/gcore_coredump.c b/src/libgcore/gcore_coredump.c
index c14cc11..424b0a4 100644
--- a/src/libgcore/gcore_coredump.c
+++ b/src/libgcore/gcore_coredump.c
@@ -78,7 +78,8 @@ readswap(ulonglong pte_val, char *buf, ulong len, ulonglong vaddr)
void gcore_readmem_user(ulong addr, void *buf, long size, char *type)
{
- ulong paddr, cnt;
+ physaddr_t paddr;
+ ulong cnt;
char *bufptr = buf;
while (size > 0) {
--
2.28.0
4:15 a.m.
New subject: [PATCH crash-gcore 2/2] gcore: fix offset types
Use loff_t consistently to fix these warnings on -m32 builds on 64-bit:
libgcore/gcore_coredump.c: In function ‘writenote’:
libgcore/gcore_coredump.c:701:58: warning: passing argument 3 of
‘gcore->elf->ops->write_note_header’ from incompatible pointer type
[-Wincompatible-pointer-types]
if (!gcore->elf->ops->write_note_header(gcore->elf, fp, foffset))
^~~~~~~
libgcore/gcore_coredump.c:701:58: note: expected ‘off_t *’ {aka ‘long
int *’} but argument is of type ‘loff_t *’ {aka ‘long long int *’}
Signed-off-by: Vincent Whitchurch <vincent.whitchurch(a)axis.com>
---
src/libgcore/gcore_defs.h | 2 +-
src/libgcore/gcore_elf_struct.c | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/src/libgcore/gcore_defs.h b/src/libgcore/gcore_defs.h
index 3233ea5..409678e 100644
--- a/src/libgcore/gcore_defs.h
+++ b/src/libgcore/gcore_defs.h
@@ -1232,7 +1232,7 @@ struct gcore_elf_operations
int (*write_section_header)(struct gcore_elf_struct *this, FILE *fp);
int (*write_program_header)(struct gcore_elf_struct *this, FILE *fp);
int (*write_note_header)(struct gcore_elf_struct *this, FILE *fp,
- off_t *offset);
+ loff_t *offset);
uint64_t (*get_e_phoff)(struct gcore_elf_struct *this);
uint64_t (*get_e_shoff)(struct gcore_elf_struct *this);
diff --git a/src/libgcore/gcore_elf_struct.c b/src/libgcore/gcore_elf_struct.c
index 2aca984..b31388a 100644
--- a/src/libgcore/gcore_elf_struct.c
+++ b/src/libgcore/gcore_elf_struct.c
@@ -141,7 +141,7 @@ static int elf64_write_program_header(struct gcore_elf_struct *this,
FILE *fp)
}
static int elf64_write_note_header(struct gcore_elf_struct *this, FILE *fp,
- off_t *offset)
+ loff_t *offset)
{
Elf64_Nhdr *n = &((struct gcore_elf64_struct *)this)->nhdr;
@@ -314,7 +314,7 @@ static int elf32_write_program_header(struct gcore_elf_struct *this,
FILE *fp)
}
static int elf32_write_note_header(struct gcore_elf_struct *this, FILE *fp,
- off_t *offset)
+ loff_t *offset)
{
Elf32_Nhdr *n = &((struct gcore_elf32_struct *)this)->nhdr;
--
2.28.0
4:52 a.m.
Hi,
Thank you for these patches.
I'm a bit tied up at the moment, but I'll review this until next week.
Please wait for a while.
________________________________________
From: Vincent Whitchurch <vincent.whitchurch(a)axis.com>
Sent: Tuesday, June 21, 2022 18:15
To: crash-utility(a)redhat.com; Hatayama, Daisuke/畑山 大輔
Cc: Vincent Whitchurch
Subject: [PATCH crash-gcore 0/2] Fixes for 32-bit ARM
This series fixes some build warnings seen when building gcore for analysis of
32-bit ARM dumps on a 64-bit x86 host. Without the fixes, gcore.so sometimes
segfaults due to stack corruption due to the type mismatches.
Thank you.
Vincent Whitchurch (2):
gcore: fix phys addr type
gcore: fix offset types
src/libgcore/gcore_coredump.c | 3 ++-
src/libgcore/gcore_defs.h | 2 +-
src/libgcore/gcore_elf_struct.c | 4 ++--
3 files changed, 5 insertions(+), 4 deletions(-)
--
2.28.0
2:52 a.m.
Hi,
Thanks for the patch set.
I merged them.
https://github.com/fujitsu/crash-gcore/commit/8ff3de974aa9fdf8934797122dc...
https://github.com/fujitsu/crash-gcore/commit/1ba701c1d7bd94cc5a02f516527...
Note that I edited their commit descriptions. In particular, for the first one, it's
segmentation fault issue rather than output of warning message.
1035
days inactive
1043
days old
devel@lists.crash-utility.osci.io
4 comments
2 participants
participants (2)
-
d.hatayama＠fujitsu.com -
Vincent Whitchurch