On 6/24/20 9:06 AM, Arnaldo Carvalho de Melo wrote:
Em Wed, Jun 24, 2020 at 12:05:50PM +0100, Lorenz Bauer escreveu:
Hi,
If pahole -J is used on an ELF that has BTF info from clang, it
produces an invalid
output. This is because pahole rewrites the .BTF section (which
includes a new string
table) but it doesn't touch .BTF.ext at all.
To demonstrate, on a recent check out of bpf-next:
$ cp connect4_prog.o connect4_pahole.o
$ pahole -J connect4_pahole.o
$ llvm-objcopy-10 --dump-section .BTF=pahole-btf.bin
--dump-section .BTF.ext=pahole-btf-ext.bin connect4_pahole.o
$ llvm-objcopy-10 --dump-section .BTF=btf.bin --dump-section
.BTF.ext=btf-ext.bin connect4_prog.o
$ sha1sum *.bin
1b5c7407dd9fd13f969931d32f6b864849e66a68 btf.bin
4c43efcc86d3cd908ddc77c15fc4a35af38d842b btf-ext.bin
2a60767a3a037de66a8d963110601769fa0f198e pahole-btf.bin
4c43efcc86d3cd908ddc77c15fc4a35af38d842b pahole-btf-ext.bin
This problem crops up when compiling old kernels like 4.19 which have
an extra pahole
build step with clang-10.
I think a possible fix is to strip .BTF.ext if .BTF is rewritten.
Agreed.
Longer term pahole needs to generate the .BTF.ext from DWARF, but then,
if clang is generating it already, why use pahole -J?
Does clang do deduplication for multi-object binaries?
No. It does not. clang did not do static linking for bpf objects
currently but we may do it in the future...
Also its nice to see that the BTF generated ends up with the same
sha1sum, cool :-)
Best
Lorenz
--
Lorenz Bauer | Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK
https://urldefense.proofpoint.com/v2/url?u=http-3A__www.cloudflare.com&d=DwIBAg&c=5VD0RTtNlTh3ycd41b3MUw&r=DA8e1B5r073vIqRrFz7MRA&m=VBET_4Cak79EDDrdr9dfzHXXwBqdhb1fmqug2a7lpPc&s=9Wa1gFDt7uR6HK2w9FgzzLXJ2pMq8Ep0i9IXaaqUnb0&e=
