from:"Artur Paszkiewicz"

[PATCH] mem: allow using ASan in multi-process mode

2023-10-04 Thread Artur Paszkiewicz

is marked as free in the shadow, accessing this address will cause an ASan error. Signed-off-by: Artur Paszkiewicz --- lib/eal/common/eal_common_memory.c | 10 lib/eal/common/eal_private.h | 22 lib/eal/linux/eal_memalloc.c | 9 +++- lib/eal/linux/eal_memory.c

[PATCH v2] mem: allow using ASan in multi-process mode

2023-10-09 Thread Artur Paszkiewicz

is marked as free in the shadow, accessing this address will cause an ASan error. Signed-off-by: Artur Paszkiewicz --- v2: - Added checks for config options disabling multi-process support. - Fixed missing unmap in legacy mode. lib/eal/common/eal_common_memory.c | 9 +++ lib/eal/common

Re: [PATCH] mem: allow using ASan in multi-process mode

2023-10-09 Thread Artur Paszkiewicz

On 10/4/23 16:51, David Marchand wrote: - did you test with --in-memory mode? with --no-huge? Please see v2 of the patch. I added checks for these options. They imply no multi-process support so mapping is skipped for those cases. - I did not look at the patch, but I wonder if there is a risk

[PATCH v3] mem: allow using ASan in multi-process mode

2023-10-25 Thread Artur Paszkiewicz

is marked as free in the shadow, accessing this address will cause an ASan error. Signed-off-by: Artur Paszkiewicz --- v3: - Removed conditional compilation from eal_common_memory.c. - Improved comments. v2: - Added checks for config options disabling multi-process support. - Fixed missing unmap

[PATCH] malloc: fix allocation for a specific case with ASAN

2023-09-04 Thread Artur Paszkiewicz

h such alignment and MALLOC_ELEM_TRAILER_LEN > 0, as correcly checked by malloc_elem_can_hold(). Signed-off-by: Artur Paszkiewicz --- lib/eal/common/malloc_heap.c | 4 ++-- lib/eal/common/malloc_mp.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/lib/eal/common/mall

[PATCH v2] malloc: fix allocation for a specific case with ASAN

2023-09-11 Thread Artur Paszkiewicz

h such alignment and MALLOC_ELEM_TRAILER_LEN > 0, as correctly checked by malloc_elem_can_hold(). Signed-off-by: Artur Paszkiewicz --- v2: - fix commit message typo lib/eal/common/malloc_heap.c | 4 ++-- lib/eal/common/malloc_mp.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions

Re: [PATCH v3] mem: allow using ASan in multi-process mode

2024-10-17 Thread Artur Paszkiewicz

On 10/3/24 23:18, Stephen Hemminger wrote: Makes sense, but patch has some fuzz against current main branch. There is also another patch that address the ASAN touch issue. https://patchwork.dpdk.org/project/dpdk/patch/20240723083419.12435-1-amic...@kalrayinc.com/ I just sent a new version of t

[PATCH RESEND] malloc: fix allocation for a specific case with ASan

2024-10-17 Thread Artur Paszkiewicz

h such alignment and MALLOC_ELEM_TRAILER_LEN > 0, as correctly checked by malloc_elem_can_hold(). Signed-off-by: Artur Paszkiewicz --- lib/eal/common/malloc_heap.c | 4 ++-- lib/eal/common/malloc_mp.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/lib/eal/common/mall

[PATCH v4] mem: allow using ASan in multi-process mode

2024-10-17 Thread Artur Paszkiewicz

. Fix it by mapping the shadow regions for allocated segments as shared memory. The primary process is responsible for creating and removing the shared memory objects. Signed-off-by: Artur Paszkiewicz --- v4: - Map ASan shadow shm after mapping the segment. Due to a change in ASan behavior[1] the