On Fri, Mar 14, 2025 at 02:40:39PM +0800, Chen Linxuan wrote: > [ Upstream commit 5ac9b4e935dfc6af41eee2ddc21deb5c36507a9f ] > > >>From memfd_secret(2) manpage: > > The memory areas backing the file created with memfd_secret(2) are > visible only to the processes that have access to the file descriptor. > The memory region is removed from the kernel page tables and only the > page tables of the processes holding the file descriptor map the > corresponding physical memory. (Thus, the pages in the region can't be > accessed by the kernel itself, so that, for example, pointers to the > region can't be passed to system calls.) > > We need to handle this special case gracefully in build ID fetching > code. Return -EFAULT whenever secretmem file is passed to build_id_parse() > family of APIs. Original report and repro can be found in [0]. > > [0] https://lore.kernel.org/bpf/ZwyG8Uro%2FSyTXAni@ly-workstation/ > > Fixes: de3ec364c3c3 ("lib/buildid: add single folio-based file reader abstraction") > Reported-by: Yi Lai <yi1.lai@xxxxxxxxx> > Suggested-by: Shakeel Butt <shakeel.butt@xxxxxxxxx> > Signed-off-by: Andrii Nakryiko <andrii@xxxxxxxxxx> > Signed-off-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx> > Acked-by: Shakeel Butt <shakeel.butt@xxxxxxxxx> > Link: https://lore.kernel.org/bpf/20241017175431.6183-A-hca@xxxxxxxxxxxxx > Link: https://lore.kernel.org/bpf/20241017174713.2157873-1-andrii@xxxxxxxxxx > [ Linxuan: perform an equivalent direct check without folio-based changes ] > Fixes: 88a16a130933 ("perf: Add build id data in mmap2 event") > Signed-off-by: Chen Linxuan <chenlinxuan@xxxxxxxxxx> > --- > > Some previous discussions can be found in the following links: > https://lore.kernel.org/stable/05D0A9F7DE394601+20250311100555.310788-2-chenlinxuan@xxxxxxxxxx/ > > --- > lib/buildid.c | 7 +++++++ > 1 file changed, 7 insertions(+) > > diff --git a/lib/buildid.c b/lib/buildid.c > index 9fc46366597e..6249bd47fb0b 100644 > --- a/lib/buildid.c > +++ b/lib/buildid.c > @@ -5,6 +5,7 @@ > #include <linux/elf.h> > #include <linux/kernel.h> > #include <linux/pagemap.h> > +#include <linux/secretmem.h> > > #define BUILD_ID 3 > > @@ -157,6 +158,12 @@ int build_id_parse(struct vm_area_struct *vma, unsigned char *build_id, > if (!vma->vm_file) > return -EINVAL; > > +#ifdef CONFIG_SECRETMEM > + /* reject secretmem folios created with memfd_secret() */ > + if (vma->vm_file->f_mapping->a_ops == &secretmem_aops) > + return -EFAULT; > +#endif We really do not like #ifdef in .c files. Why can't this use much the same call it does in the original commit? Just put that in the correct .h file so that the #ifdef is not needed here, to make the backport match much more cleanly. thanks, greg k-h
