Message ID | 12cec1262be71de5f1d9eae121b637041a5ae247.1723459079.git.sam@gentoo.org (mailing list archive) |
---|---|
State | Superseded |
Delegated to: | BPF |
Headers | show |
Series | [v3] libbpf: workaround -Wmaybe-uninitialized false positive | expand |
On Mon, Aug 12, 2024 at 11:37:59AM +0100, Sam James wrote: > In `elf_close`, we get this with GCC 15 -O3 (at least): > ``` > In function ‘elf_close’, > inlined from ‘elf_close’ at elf.c:53:6, > inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: > elf.c:57:9: warning: ‘elf_fd.elf’ may be used uninitialized [-Wmaybe-uninitialized] > 57 | elf_end(elf_fd->elf); > | ^~~~~~~~~~~~~~~~~~~~ > elf.c: In function ‘elf_find_func_offset_from_file’: > elf.c:377:23: note: ‘elf_fd.elf’ was declared here > 377 | struct elf_fd elf_fd; > | ^~~~~~ > In function ‘elf_close’, > inlined from ‘elf_close’ at elf.c:53:6, > inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: > elf.c:58:9: warning: ‘elf_fd.fd’ may be used uninitialized [-Wmaybe-uninitialized] > 58 | close(elf_fd->fd); > | ^~~~~~~~~~~~~~~~~ > elf.c: In function ‘elf_find_func_offset_from_file’: > elf.c:377:23: note: ‘elf_fd.fd’ was declared here > 377 | struct elf_fd elf_fd; > | ^~~~~~ > ``` > > In reality, our use is fine, it's just that GCC doesn't model errno > here (see linked GCC bug). Suppress -Wmaybe-uninitialized accordingly > by initializing elf_fd.elf to -1. > > I've done this in two other functions as well given it could easily > occur there too (same access/use pattern). > > Link: https://gcc.gnu.org/PR114952 > Signed-off-by: Sam James <sam@gentoo.org> > --- > v3: Initialize to -1 instead of using a pragma. it's false positive, but I wonder we could still add Fixes tag Acked-by: Jiri Olsa <jolsa@kernel.org> jirka > > Range-diff against v2: > 1: 8f5c3b173e4cb < -: ------------- libbpf: workaround -Wmaybe-uninitialized false positive > -: ------------- > 1: 12cec1262be71 libbpf: workaround -Wmaybe-uninitialized false positive > > tools/lib/bpf/elf.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/tools/lib/bpf/elf.c b/tools/lib/bpf/elf.c > index c92e02394159e..00ea3f867bbc8 100644 > --- a/tools/lib/bpf/elf.c > +++ b/tools/lib/bpf/elf.c > @@ -374,7 +374,7 @@ long elf_find_func_offset(Elf *elf, const char *binary_path, const char *name) > */ > long elf_find_func_offset_from_file(const char *binary_path, const char *name) > { > - struct elf_fd elf_fd; > + struct elf_fd elf_fd = { .fd = -1 }; > long ret = -ENOENT; > > ret = elf_open(binary_path, &elf_fd); > @@ -412,7 +412,7 @@ int elf_resolve_syms_offsets(const char *binary_path, int cnt, > int err = 0, i, cnt_done = 0; > unsigned long *offsets; > struct symbol *symbols; > - struct elf_fd elf_fd; > + struct elf_fd elf_fd = { .fd = -1 }; > > err = elf_open(binary_path, &elf_fd); > if (err) > @@ -507,7 +507,7 @@ int elf_resolve_pattern_offsets(const char *binary_path, const char *pattern, > int sh_types[2] = { SHT_SYMTAB, SHT_DYNSYM }; > unsigned long *offsets = NULL; > size_t cap = 0, cnt = 0; > - struct elf_fd elf_fd; > + struct elf_fd elf_fd = { .fd = -1 }; > int err = 0, i; > > err = elf_open(binary_path, &elf_fd); > -- > 2.45.2 >
On 12/08/2024 11:37, Sam James wrote: > In `elf_close`, we get this with GCC 15 -O3 (at least): > ``` > In function ‘elf_close’, > inlined from ‘elf_close’ at elf.c:53:6, > inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: > elf.c:57:9: warning: ‘elf_fd.elf’ may be used uninitialized [-Wmaybe-uninitialized] > 57 | elf_end(elf_fd->elf); > | ^~~~~~~~~~~~~~~~~~~~ > elf.c: In function ‘elf_find_func_offset_from_file’: > elf.c:377:23: note: ‘elf_fd.elf’ was declared here > 377 | struct elf_fd elf_fd; > | ^~~~~~ > In function ‘elf_close’, > inlined from ‘elf_close’ at elf.c:53:6, > inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: > elf.c:58:9: warning: ‘elf_fd.fd’ may be used uninitialized [-Wmaybe-uninitialized] > 58 | close(elf_fd->fd); > | ^~~~~~~~~~~~~~~~~ > elf.c: In function ‘elf_find_func_offset_from_file’: > elf.c:377:23: note: ‘elf_fd.fd’ was declared here > 377 | struct elf_fd elf_fd; > | ^~~~~~ > ``` > > In reality, our use is fine, it's just that GCC doesn't model errno > here (see linked GCC bug). Suppress -Wmaybe-uninitialized accordingly > by initializing elf_fd.elf to -1. > > I've done this in two other functions as well given it could easily > occur there too (same access/use pattern). > hmm, looking at this again - given that there are multiple consumers - I suppose another option would perhaps be to - have elf_open() to init int fd = -1, Elf *elf = NULL. - have error paths in elf_open() "goto out"; at out: we set elf_fd->fd, elf_fd->elf to fd, elf - have elf_close() exit it elf_fd < 0 (since 0 is a valid fd), as it will for the error cases Might all be bit excessive, and might not even fix the false positive issue here, so > Link: https://gcc.gnu.org/PR114952 > Signed-off-by: Sam James <sam@gentoo.org> Reviewed-by: Alan Maguire <alan.maguire@oracle.com> > --- > v3: Initialize to -1 instead of using a pragma. > > Range-diff against v2: > 1: 8f5c3b173e4cb < -: ------------- libbpf: workaround -Wmaybe-uninitialized false positive > -: ------------- > 1: 12cec1262be71 libbpf: workaround -Wmaybe-uninitialized false positive > > tools/lib/bpf/elf.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/tools/lib/bpf/elf.c b/tools/lib/bpf/elf.c > index c92e02394159e..00ea3f867bbc8 100644 > --- a/tools/lib/bpf/elf.c > +++ b/tools/lib/bpf/elf.c > @@ -374,7 +374,7 @@ long elf_find_func_offset(Elf *elf, const char *binary_path, const char *name) > */ > long elf_find_func_offset_from_file(const char *binary_path, const char *name) > { > - struct elf_fd elf_fd; > + struct elf_fd elf_fd = { .fd = -1 }; > long ret = -ENOENT; > > ret = elf_open(binary_path, &elf_fd); > @@ -412,7 +412,7 @@ int elf_resolve_syms_offsets(const char *binary_path, int cnt, > int err = 0, i, cnt_done = 0; > unsigned long *offsets; > struct symbol *symbols; > - struct elf_fd elf_fd; > + struct elf_fd elf_fd = { .fd = -1 }; > > err = elf_open(binary_path, &elf_fd); > if (err) > @@ -507,7 +507,7 @@ int elf_resolve_pattern_offsets(const char *binary_path, const char *pattern, > int sh_types[2] = { SHT_SYMTAB, SHT_DYNSYM }; > unsigned long *offsets = NULL; > size_t cap = 0, cnt = 0; > - struct elf_fd elf_fd; > + struct elf_fd elf_fd = { .fd = -1 }; > int err = 0, i; > > err = elf_open(binary_path, &elf_fd);
On Mon, Aug 12, 2024 at 6:57 AM Alan Maguire <alan.maguire@oracle.com> wrote: > > On 12/08/2024 11:37, Sam James wrote: > > In `elf_close`, we get this with GCC 15 -O3 (at least): > > ``` > > In function ‘elf_close’, > > inlined from ‘elf_close’ at elf.c:53:6, > > inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: > > elf.c:57:9: warning: ‘elf_fd.elf’ may be used uninitialized [-Wmaybe-uninitialized] > > 57 | elf_end(elf_fd->elf); > > | ^~~~~~~~~~~~~~~~~~~~ > > elf.c: In function ‘elf_find_func_offset_from_file’: > > elf.c:377:23: note: ‘elf_fd.elf’ was declared here > > 377 | struct elf_fd elf_fd; > > | ^~~~~~ > > In function ‘elf_close’, > > inlined from ‘elf_close’ at elf.c:53:6, > > inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: > > elf.c:58:9: warning: ‘elf_fd.fd’ may be used uninitialized [-Wmaybe-uninitialized] > > 58 | close(elf_fd->fd); > > | ^~~~~~~~~~~~~~~~~ > > elf.c: In function ‘elf_find_func_offset_from_file’: > > elf.c:377:23: note: ‘elf_fd.fd’ was declared here > > 377 | struct elf_fd elf_fd; > > | ^~~~~~ > > ``` > > > > In reality, our use is fine, it's just that GCC doesn't model errno > > here (see linked GCC bug). Suppress -Wmaybe-uninitialized accordingly > > by initializing elf_fd.elf to -1. > > > > I've done this in two other functions as well given it could easily > > occur there too (same access/use pattern). > > > > hmm, looking at this again - given that there are multiple consumers - yes, I don't like that each caller has to remember to initialize the struct that is clearly initialized by elf_open() itself, so see below. pw-bot: cr > I suppose another option would perhaps be to > > - have elf_open() to init int fd = -1, Elf *elf = NULL. I'd do just elf_fd->elf = NULL; elf_fd->fd = -1; and do nothing else. This should be enough for compiler to not trigger this. > - have error paths in elf_open() "goto out"; at out: we set elf_fd->fd, > elf_fd->elf to fd, elf > - have elf_close() exit it elf_fd < 0 (since 0 is a valid fd), as it > will for the error cases > Let's not touch anything else, this should be enough. > Might all be bit excessive, and might not even fix the false positive > issue here, so > > > Link: https://gcc.gnu.org/PR114952 > > Signed-off-by: Sam James <sam@gentoo.org> > > Reviewed-by: Alan Maguire <alan.maguire@oracle.com> > > > --- > > v3: Initialize to -1 instead of using a pragma. > > > > Range-diff against v2: > > 1: 8f5c3b173e4cb < -: ------------- libbpf: workaround -Wmaybe-uninitialized false positive > > -: ------------- > 1: 12cec1262be71 libbpf: workaround -Wmaybe-uninitialized false positive > > > > tools/lib/bpf/elf.c | 6 +++--- > > 1 file changed, 3 insertions(+), 3 deletions(-) > > > > diff --git a/tools/lib/bpf/elf.c b/tools/lib/bpf/elf.c > > index c92e02394159e..00ea3f867bbc8 100644 > > --- a/tools/lib/bpf/elf.c > > +++ b/tools/lib/bpf/elf.c > > @@ -374,7 +374,7 @@ long elf_find_func_offset(Elf *elf, const char *binary_path, const char *name) > > */ > > long elf_find_func_offset_from_file(const char *binary_path, const char *name) > > { > > - struct elf_fd elf_fd; > > + struct elf_fd elf_fd = { .fd = -1 }; > > long ret = -ENOENT; > > > > ret = elf_open(binary_path, &elf_fd); > > @@ -412,7 +412,7 @@ int elf_resolve_syms_offsets(const char *binary_path, int cnt, > > int err = 0, i, cnt_done = 0; > > unsigned long *offsets; > > struct symbol *symbols; > > - struct elf_fd elf_fd; > > + struct elf_fd elf_fd = { .fd = -1 }; > > > > err = elf_open(binary_path, &elf_fd); > > if (err) > > @@ -507,7 +507,7 @@ int elf_resolve_pattern_offsets(const char *binary_path, const char *pattern, > > int sh_types[2] = { SHT_SYMTAB, SHT_DYNSYM }; > > unsigned long *offsets = NULL; > > size_t cap = 0, cnt = 0; > > - struct elf_fd elf_fd; > > + struct elf_fd elf_fd = { .fd = -1 }; > > int err = 0, i; > > > > err = elf_open(binary_path, &elf_fd);
Andrii Nakryiko <andrii.nakryiko@gmail.com> writes: > On Mon, Aug 12, 2024 at 6:57 AM Alan Maguire <alan.maguire@oracle.com> wrote: >> >> On 12/08/2024 11:37, Sam James wrote: >> > In `elf_close`, we get this with GCC 15 -O3 (at least): >> > ``` >> > In function ‘elf_close’, >> > inlined from ‘elf_close’ at elf.c:53:6, >> > inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: >> > elf.c:57:9: warning: ‘elf_fd.elf’ may be used uninitialized [-Wmaybe-uninitialized] >> > 57 | elf_end(elf_fd->elf); >> > | ^~~~~~~~~~~~~~~~~~~~ >> > elf.c: In function ‘elf_find_func_offset_from_file’: >> > elf.c:377:23: note: ‘elf_fd.elf’ was declared here >> > 377 | struct elf_fd elf_fd; >> > | ^~~~~~ >> > In function ‘elf_close’, >> > inlined from ‘elf_close’ at elf.c:53:6, >> > inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: >> > elf.c:58:9: warning: ‘elf_fd.fd’ may be used uninitialized [-Wmaybe-uninitialized] >> > 58 | close(elf_fd->fd); >> > | ^~~~~~~~~~~~~~~~~ >> > elf.c: In function ‘elf_find_func_offset_from_file’: >> > elf.c:377:23: note: ‘elf_fd.fd’ was declared here >> > 377 | struct elf_fd elf_fd; >> > | ^~~~~~ >> > ``` >> > >> > In reality, our use is fine, it's just that GCC doesn't model errno >> > here (see linked GCC bug). Suppress -Wmaybe-uninitialized accordingly >> > by initializing elf_fd.elf to -1. >> > >> > I've done this in two other functions as well given it could easily >> > occur there too (same access/use pattern). >> > >> >> hmm, looking at this again - given that there are multiple consumers - > > yes, I don't like that each caller has to remember to initialize the > struct that is clearly initialized by elf_open() itself, so see below. > > pw-bot: cr > >> I suppose another option would perhaps be to >> >> - have elf_open() to init int fd = -1, Elf *elf = NULL. > > I'd do just > > elf_fd->elf = NULL; > elf_fd->fd = -1; > > and do nothing else. This should be enough for compiler to not trigger this. OK. > >> - have error paths in elf_open() "goto out"; at out: we set elf_fd->fd, >> elf_fd->elf to fd, elf >> - have elf_close() exit it elf_fd < 0 (since 0 is a valid fd), as it >> will for the error cases >> > > Let's not touch anything else, this should be enough. > > >> Might all be bit excessive, and might not even fix the false positive >> issue here, so >> >> > Link: https://gcc.gnu.org/PR114952 >> > Signed-off-by: Sam James <sam@gentoo.org> >> >> Reviewed-by: Alan Maguire <alan.maguire@oracle.com> >> >> > --- >> > v3: Initialize to -1 instead of using a pragma. >> > >> > Range-diff against v2: >> > 1: 8f5c3b173e4cb < -: ------------- libbpf: workaround -Wmaybe-uninitialized false positive >> > -: ------------- > 1: 12cec1262be71 libbpf: workaround -Wmaybe-uninitialized false positive >> > >> > tools/lib/bpf/elf.c | 6 +++--- >> > 1 file changed, 3 insertions(+), 3 deletions(-) >> > >> > diff --git a/tools/lib/bpf/elf.c b/tools/lib/bpf/elf.c >> > index c92e02394159e..00ea3f867bbc8 100644 >> > --- a/tools/lib/bpf/elf.c >> > +++ b/tools/lib/bpf/elf.c >> > @@ -374,7 +374,7 @@ long elf_find_func_offset(Elf *elf, const char *binary_path, const char *name) >> > */ >> > long elf_find_func_offset_from_file(const char *binary_path, const char *name) >> > { >> > - struct elf_fd elf_fd; >> > + struct elf_fd elf_fd = { .fd = -1 }; >> > long ret = -ENOENT; >> > >> > ret = elf_open(binary_path, &elf_fd); >> > @@ -412,7 +412,7 @@ int elf_resolve_syms_offsets(const char *binary_path, int cnt, >> > int err = 0, i, cnt_done = 0; >> > unsigned long *offsets; >> > struct symbol *symbols; >> > - struct elf_fd elf_fd; >> > + struct elf_fd elf_fd = { .fd = -1 }; >> > >> > err = elf_open(binary_path, &elf_fd); >> > if (err) >> > @@ -507,7 +507,7 @@ int elf_resolve_pattern_offsets(const char *binary_path, const char *pattern, >> > int sh_types[2] = { SHT_SYMTAB, SHT_DYNSYM }; >> > unsigned long *offsets = NULL; >> > size_t cap = 0, cnt = 0; >> > - struct elf_fd elf_fd; >> > + struct elf_fd elf_fd = { .fd = -1 }; >> > int err = 0, i; >> > >> > err = elf_open(binary_path, &elf_fd);
diff --git a/tools/lib/bpf/elf.c b/tools/lib/bpf/elf.c index c92e02394159e..00ea3f867bbc8 100644 --- a/tools/lib/bpf/elf.c +++ b/tools/lib/bpf/elf.c @@ -374,7 +374,7 @@ long elf_find_func_offset(Elf *elf, const char *binary_path, const char *name) */ long elf_find_func_offset_from_file(const char *binary_path, const char *name) { - struct elf_fd elf_fd; + struct elf_fd elf_fd = { .fd = -1 }; long ret = -ENOENT; ret = elf_open(binary_path, &elf_fd); @@ -412,7 +412,7 @@ int elf_resolve_syms_offsets(const char *binary_path, int cnt, int err = 0, i, cnt_done = 0; unsigned long *offsets; struct symbol *symbols; - struct elf_fd elf_fd; + struct elf_fd elf_fd = { .fd = -1 }; err = elf_open(binary_path, &elf_fd); if (err) @@ -507,7 +507,7 @@ int elf_resolve_pattern_offsets(const char *binary_path, const char *pattern, int sh_types[2] = { SHT_SYMTAB, SHT_DYNSYM }; unsigned long *offsets = NULL; size_t cap = 0, cnt = 0; - struct elf_fd elf_fd; + struct elf_fd elf_fd = { .fd = -1 }; int err = 0, i; err = elf_open(binary_path, &elf_fd);
In `elf_close`, we get this with GCC 15 -O3 (at least): ``` In function ‘elf_close’, inlined from ‘elf_close’ at elf.c:53:6, inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: elf.c:57:9: warning: ‘elf_fd.elf’ may be used uninitialized [-Wmaybe-uninitialized] 57 | elf_end(elf_fd->elf); | ^~~~~~~~~~~~~~~~~~~~ elf.c: In function ‘elf_find_func_offset_from_file’: elf.c:377:23: note: ‘elf_fd.elf’ was declared here 377 | struct elf_fd elf_fd; | ^~~~~~ In function ‘elf_close’, inlined from ‘elf_close’ at elf.c:53:6, inlined from ‘elf_find_func_offset_from_file’ at elf.c:384:2: elf.c:58:9: warning: ‘elf_fd.fd’ may be used uninitialized [-Wmaybe-uninitialized] 58 | close(elf_fd->fd); | ^~~~~~~~~~~~~~~~~ elf.c: In function ‘elf_find_func_offset_from_file’: elf.c:377:23: note: ‘elf_fd.fd’ was declared here 377 | struct elf_fd elf_fd; | ^~~~~~ ``` In reality, our use is fine, it's just that GCC doesn't model errno here (see linked GCC bug). Suppress -Wmaybe-uninitialized accordingly by initializing elf_fd.elf to -1. I've done this in two other functions as well given it could easily occur there too (same access/use pattern). Link: https://gcc.gnu.org/PR114952 Signed-off-by: Sam James <sam@gentoo.org> --- v3: Initialize to -1 instead of using a pragma. Range-diff against v2: 1: 8f5c3b173e4cb < -: ------------- libbpf: workaround -Wmaybe-uninitialized false positive -: ------------- > 1: 12cec1262be71 libbpf: workaround -Wmaybe-uninitialized false positive tools/lib/bpf/elf.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)