| Message ID | 20241021014004.1647816-1-houtao@huaweicloud.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Misc fixes for bpf | expand |
On Sun, Oct 20, 2024 at 6:28 PM Hou Tao <houtao@huaweicloud.com> wrote: > > From: Hou Tao <houtao1@huawei.com> > > Hi, > > The patch set is just a bundle of fixes. Patch #1 fixes the out-of-bound > for sockmap link fdinfo. Patch #2 adds a static assertion to guarantee > such out-of-bound access will never happen again. Patch #3 fixes the > kmemleak report when parsing the mount options for bpffs. Patch #4~#7 > fix issues related to bits iterator. > > Please check the individual patches for more details. And comments are > always welcome. > > v2: > * patch #1: update tools/include/uapi/linux/bpf.h as well (Alexei) > * patch #2: new patch. Add a static assertion for bpf_link_type_strs[] (Andrii) > * patch #4: doesn't set nr_bits to zero in bpf_iter_bits_next (Andrii) > * patch #5: use 512 as the maximal value of nr_words > * patch #6: change the type of both bits and bits_copy to u64 (Andrii) > > v1: https://lore.kernel.org/bpf/20241008091718.3797027-1-houtao@huaweicloud.com/ > You have three separate groups of fixes, please send them separately so they can be landed separately and tested separately: > Hou Tao (7): > bpf: Add the missing BPF_LINK_TYPE invocation for sockmap > bpf: Add assertion for the size of bpf_link_type_strs[] first, link type fixes > bpf: Preserve param->string when parsing mount options parsing fix > bpf: Free dynamically allocated bits in bpf_iter_bits_destroy() > bpf: Check the validity of nr_words in bpf_iter_bits_new() > bpf: Use __u64 to save the bits in bits iterator > selftests/bpf: Test multiplication overflow of nr_bits in bits_iter bits iter fixes > > include/linux/bpf_types.h | 7 +-- > include/uapi/linux/bpf.h | 3 ++ > kernel/bpf/helpers.c | 45 +++++++++++++++---- > kernel/bpf/inode.c | 5 ++- > kernel/bpf/syscall.c | 2 + > tools/include/uapi/linux/bpf.h | 3 ++ > .../selftests/bpf/progs/verifier_bits_iter.c | 14 ++++++ > 7 files changed, 63 insertions(+), 16 deletions(-) > > -- > 2.29.2 >
Hi Andrii, On 10/22/2024 7:11 AM, Andrii Nakryiko wrote: > On Sun, Oct 20, 2024 at 6:28 PM Hou Tao <houtao@huaweicloud.com> wrote: >> From: Hou Tao <houtao1@huawei.com> >> >> Hi, >> >> The patch set is just a bundle of fixes. Patch #1 fixes the out-of-bound >> for sockmap link fdinfo. Patch #2 adds a static assertion to guarantee >> such out-of-bound access will never happen again. Patch #3 fixes the >> kmemleak report when parsing the mount options for bpffs. Patch #4~#7 >> fix issues related to bits iterator. >> >> Please check the individual patches for more details. And comments are >> always welcome. >> >> v2: >> * patch #1: update tools/include/uapi/linux/bpf.h as well (Alexei) >> * patch #2: new patch. Add a static assertion for bpf_link_type_strs[] (Andrii) >> * patch #4: doesn't set nr_bits to zero in bpf_iter_bits_next (Andrii) >> * patch #5: use 512 as the maximal value of nr_words >> * patch #6: change the type of both bits and bits_copy to u64 (Andrii) >> >> v1: https://lore.kernel.org/bpf/20241008091718.3797027-1-houtao@huaweicloud.com/ >> > You have three separate groups of fixes, please send them separately > so they can be landed separately and tested separately: > >> Hou Tao (7): >> bpf: Add the missing BPF_LINK_TYPE invocation for sockmap >> bpf: Add assertion for the size of bpf_link_type_strs[] > first, link type fixes > >> bpf: Preserve param->string when parsing mount options > parsing fix > >> bpf: Free dynamically allocated bits in bpf_iter_bits_destroy() >> bpf: Check the validity of nr_words in bpf_iter_bits_new() >> bpf: Use __u64 to save the bits in bits iterator >> selftests/bpf: Test multiplication overflow of nr_bits in bits_iter > bits iter fixes Thanks for the suggestion. Will do in v3.
From: Hou Tao <houtao1@huawei.com> Hi, The patch set is just a bundle of fixes. Patch #1 fixes the out-of-bound for sockmap link fdinfo. Patch #2 adds a static assertion to guarantee such out-of-bound access will never happen again. Patch #3 fixes the kmemleak report when parsing the mount options for bpffs. Patch #4~#7 fix issues related to bits iterator. Please check the individual patches for more details. And comments are always welcome. v2: * patch #1: update tools/include/uapi/linux/bpf.h as well (Alexei) * patch #2: new patch. Add a static assertion for bpf_link_type_strs[] (Andrii) * patch #4: doesn't set nr_bits to zero in bpf_iter_bits_next (Andrii) * patch #5: use 512 as the maximal value of nr_words * patch #6: change the type of both bits and bits_copy to u64 (Andrii) v1: https://lore.kernel.org/bpf/20241008091718.3797027-1-houtao@huaweicloud.com/ Hou Tao (7): bpf: Add the missing BPF_LINK_TYPE invocation for sockmap bpf: Add assertion for the size of bpf_link_type_strs[] bpf: Preserve param->string when parsing mount options bpf: Free dynamically allocated bits in bpf_iter_bits_destroy() bpf: Check the validity of nr_words in bpf_iter_bits_new() bpf: Use __u64 to save the bits in bits iterator selftests/bpf: Test multiplication overflow of nr_bits in bits_iter include/linux/bpf_types.h | 7 +-- include/uapi/linux/bpf.h | 3 ++ kernel/bpf/helpers.c | 45 +++++++++++++++---- kernel/bpf/inode.c | 5 ++- kernel/bpf/syscall.c | 2 + tools/include/uapi/linux/bpf.h | 3 ++ .../selftests/bpf/progs/verifier_bits_iter.c | 14 ++++++ 7 files changed, 63 insertions(+), 16 deletions(-)