2021-06-16 22:44:06

by Mark-PK Tsai (蔡沛剛)

Subject: [PATCH v4] recordmcount: Correct st_shndx handling

From: Peter Zijlstra <[email protected]>

One should only use st_shndx when >SHN_UNDEF and <SHN_LORESERVE. When
SHN_XINDEX, then use .symtab_shndx. Otherwise use 0.

This handles the case: st_shndx >= SHN_LORESERVE && st_shndx != SHN_XINDEX.

Reported-by: Mark-PK Tsai <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Tested-by: Mark-PK Tsai <[email protected]>
[handle endianness of sym->st_shndx]
Signed-off-by: Mark-PK Tsai <[email protected]>
---
scripts/recordmcount.h | 15 ++++++++++-----
1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/scripts/recordmcount.h b/scripts/recordmcount.h
index f9b19524da11..1e9baa5c4fc6 100644
--- a/scripts/recordmcount.h
+++ b/scripts/recordmcount.h
@@ -192,15 +192,20 @@ static unsigned int get_symindex(Elf_Sym const *sym, Elf32_Word const *symtab,
Elf32_Word const *symtab_shndx)
{
unsigned long offset;
+ unsigned short shndx = w2(sym->st_shndx);
int index;

- if (sym->st_shndx != SHN_XINDEX)
- return w2(sym->st_shndx);
+ if (shndx > SHN_UNDEF && shndx < SHN_LORESERVE)
+ return shndx;

- offset = (unsigned long)sym - (unsigned long)symtab;
- index = offset / sizeof(*sym);
+ if (shndx == SHN_XINDEX) {
+ offset = (unsigned long)sym - (unsigned long)symtab;
+ index = offset / sizeof(*sym);

- return w(symtab_shndx[index]);
+ return w(symtab_shndx[index]);
+ }
+
+ return 0;
}

static unsigned int get_shnum(Elf_Ehdr const *ehdr, Elf_Shdr const *shdr0)
--
2.18.0

2021-06-16 22:50:08

by Steven Rostedt

Subject: Re: [PATCH v4] recordmcount: Correct st_shndx handling

On Wed, 16 Jun 2021 23:41:26 +0800
Mark-PK Tsai <[email protected]> wrote:

> From: Peter Zijlstra <[email protected]>
>
> One should only use st_shndx when >SHN_UNDEF and <SHN_LORESERVE. When
> SHN_XINDEX, then use .symtab_shndx. Otherwise use 0.
>
> This handles the case: st_shndx >= SHN_LORESERVE && st_shndx != SHN_XINDEX.
>
> Reported-by: Mark-PK Tsai <[email protected]>
> Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
> Tested-by: Mark-PK Tsai <[email protected]>
> [handle endianness of sym->st_shndx]
> Signed-off-by: Mark-PK Tsai <[email protected]>
> ---

Thanks, I pulled this in and started the testing.

-- Steve

2021-06-17 02:41:42

by Steven Rostedt

Subject: Re: [PATCH v4] recordmcount: Correct st_shndx handling

On Wed, 16 Jun 2021 12:02:27 -0400
Steven Rostedt <[email protected]> wrote:

> On Wed, 16 Jun 2021 23:41:26 +0800
> Mark-PK Tsai <[email protected]> wrote:
>
> > From: Peter Zijlstra <[email protected]>
> >
> > One should only use st_shndx when >SHN_UNDEF and <SHN_LORESERVE. When
> > SHN_XINDEX, then use .symtab_shndx. Otherwise use 0.
> >
> > This handles the case: st_shndx >= SHN_LORESERVE && st_shndx != SHN_XINDEX.
> >
> > Reported-by: Mark-PK Tsai <[email protected]>
> > Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
> > Tested-by: Mark-PK Tsai <[email protected]>
> > [handle endianness of sym->st_shndx]
> > Signed-off-by: Mark-PK Tsai <[email protected]>
> > ---
>
> Thanks, I pulled this in and started the testing.
>

It passed my normal tests. I'll run my cross compile test, and I just
pushed it to linux-next. I'll wait a few days before I push it to Linus.

-- Steve