linux-hardening - Re: [PATCH] Revert "bpf: Take return from set_memory_rox() into account with bpf_jit_binary_lock

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <a1dac525-4e6d-4d28-87ee-63723abbafad@cs-soprasteria.com>
Date: Mon, 8 Jul 2024 15:12:55 +0000
From: LEROY Christophe <christophe.leroy2@...soprasteria.com>
To: Greg KH <gregkh@...uxfoundation.org>, WangYuli <wangyuli@...ontech.com>
CC: "stable@...r.kernel.org" <stable@...r.kernel.org>, "sashal@...nel.org"
	<sashal@...nel.org>, "ast@...nel.org" <ast@...nel.org>,
	"keescook@...omium.org" <keescook@...omium.org>,
	"linux-hardening@...r.kernel.org" <linux-hardening@...r.kernel.org>,
	"catalin.marinas@....com" <catalin.marinas@....com>, "song@...nel.org"
	<song@...nel.org>, "puranjay12@...il.com" <puranjay12@...il.com>,
	"daniel@...earbox.net" <daniel@...earbox.net>, "andrii@...nel.org"
	<andrii@...nel.org>, "martin.lau@...ux.dev" <martin.lau@...ux.dev>,
	"yonghong.song@...ux.dev" <yonghong.song@...ux.dev>,
	"john.fastabend@...il.com" <john.fastabend@...il.com>, "kpsingh@...nel.org"
	<kpsingh@...nel.org>, "sdf@...gle.com" <sdf@...gle.com>, "haoluo@...gle.com"
	<haoluo@...gle.com>, "jolsa@...nel.org" <jolsa@...nel.org>,
	"illusionist.neo@...il.com" <illusionist.neo@...il.com>,
	"linux@...linux.org.uk" <linux@...linux.org.uk>, "bpf@...r.kernel.org"
	<bpf@...r.kernel.org>, "linux-arm-kernel@...ts.infradead.org"
	<linux-arm-kernel@...ts.infradead.org>, "linux-kernel@...r.kernel.org"
	<linux-kernel@...r.kernel.org>, "chenhuacai@...nel.org"
	<chenhuacai@...nel.org>, "kernel@...0n.name" <kernel@...0n.name>,
	"loongarch@...ts.linux.dev" <loongarch@...ts.linux.dev>,
	"johan.almbladh@...finetworks.com" <johan.almbladh@...finetworks.com>,
	"paulburton@...nel.org" <paulburton@...nel.org>, "tsbogend@...ha.franken.de"
	<tsbogend@...ha.franken.de>, "linux-mips@...r.kernel.org"
	<linux-mips@...r.kernel.org>, "deller@....de" <deller@....de>,
	"linux-parisc@...r.kernel.org" <linux-parisc@...r.kernel.org>,
	"iii@...ux.ibm.com" <iii@...ux.ibm.com>, "hca@...ux.ibm.com"
	<hca@...ux.ibm.com>, "gor@...ux.ibm.com" <gor@...ux.ibm.com>,
	"agordeev@...ux.ibm.com" <agordeev@...ux.ibm.com>,
	"borntraeger@...ux.ibm.com" <borntraeger@...ux.ibm.com>,
	"svens@...ux.ibm.com" <svens@...ux.ibm.com>, "linux-s390@...r.kernel.org"
	<linux-s390@...r.kernel.org>, "davem@...emloft.net" <davem@...emloft.net>,
	"sparclinux@...r.kernel.org" <sparclinux@...r.kernel.org>, "kuba@...nel.org"
	<kuba@...nel.org>, "hawk@...nel.org" <hawk@...nel.org>,
	"netdev@...r.kernel.org" <netdev@...r.kernel.org>, "dsahern@...nel.org"
	<dsahern@...nel.org>, "tglx@...utronix.de" <tglx@...utronix.de>,
	"mingo@...hat.com" <mingo@...hat.com>, "bp@...en8.de" <bp@...en8.de>,
	"dave.hansen@...ux.intel.com" <dave.hansen@...ux.intel.com>, "x86@...nel.org"
	<x86@...nel.org>, "hpa@...or.com" <hpa@...or.com>, "guanwentao@...ontech.com"
	<guanwentao@...ontech.com>, "baimingcong@...ontech.com"
	<baimingcong@...ontech.com>
Subject: Re: [PATCH] Revert "bpf: Take return from set_memory_rox() into
 account with bpf_jit_binary_lock_ro()" for linux-6.6.37



Le 08/07/2024 à 14:36, Greg KH a écrit :
> On Sun, Jul 07, 2024 at 03:34:15PM +0800, WangYuli wrote:
>>
>> On 2024/7/6 17:30, Greg KH wrote:
>>> This makes it sound like you are reverting this because of a build
>>> error, which is not the case here, right?  Isn't this because of the
>>> powerpc issue reported here:
>>>     https://lore.kernel.org/r/20240705203413.wbv2nw3747vjeibk@altlinux.org
>>> ?
>>
>> No, it only occurs on ARM64 architecture. The reason is that before being
>> modified, the function
>>
>> bpf_jit_binary_lock_ro() in arch/arm64/net/bpf_jit_comp.c +1651
>>
>> was introduced with __must_check, which is defined as
>> __attribute__((__warn_unused_result__)).
>>
>>
>> However, at this point, calling bpf_jit_binary_lock_ro(header)
>> coincidentally results in an unused-result
>>
>> warning.
>
> Ok, thanks, but why is no one else seeing this in their testing?

Probably the configs used by robots do not activate BPF JIT ?

>
>>> If not, why not just backport the single missing arm64 commit,
>>
>> Upstream commit 1dad391daef1 ("bpf, arm64: use bpf_prog_pack for memory
>> management") is part of
>>
>> a larger change that involves multiple commits. It's not an isolated commit.
>>
>>
>> We could certainly backport all of them to solve this problem, buthas it's not
>> the simplest solution.
>
> reverting the change feels wrong in that you will still have the bug
> present that it was trying to solve, right?  If so, can you then provide
> a working version?

Indeed, by reverting the change you "punish" all architectures because
arm64 hasn't properly been backported, is it fair ?

In fact, when I implemented commit e60adf513275 ("bpf: Take return from
set_memory_rox() into account with bpf_jit_binary_lock_ro()"), we had
the following users for function bpf_jit_binary_lock_ro() :

$ git grep bpf_jit_binary_lock_ro e60adf513275~
e60adf513275~:arch/arm/net/bpf_jit_32.c:
bpf_jit_binary_lock_ro(header);
e60adf513275~:arch/loongarch/net/bpf_jit.c:
bpf_jit_binary_lock_ro(header);
e60adf513275~:arch/mips/net/bpf_jit_comp.c:
bpf_jit_binary_lock_ro(header);
e60adf513275~:arch/parisc/net/bpf_jit_core.c:
bpf_jit_binary_lock_ro(jit_data->header);
e60adf513275~:arch/s390/net/bpf_jit_comp.c:
bpf_jit_binary_lock_ro(header);
e60adf513275~:arch/sparc/net/bpf_jit_comp_64.c:
bpf_jit_binary_lock_ro(header);
e60adf513275~:arch/x86/net/bpf_jit_comp32.c:
bpf_jit_binary_lock_ro(header);
e60adf513275~:include/linux/filter.h:static inline void
bpf_jit_binary_lock_ro(struct bpf_binary_header *hdr)

But when commit 08f6c05feb1d ("bpf: Take return from set_memory_rox()
into account with bpf_jit_binary_lock_ro()") was applied, we had one
more user which is arm64:

$ git grep bpf_jit_binary_lock_ro 08f6c05feb1d~
08f6c05feb1d~:arch/arm/net/bpf_jit_32.c:
bpf_jit_binary_lock_ro(header);
08f6c05feb1d~:arch/arm64/net/bpf_jit_comp.c:
bpf_jit_binary_lock_ro(header);
08f6c05feb1d~:arch/loongarch/net/bpf_jit.c:
bpf_jit_binary_lock_ro(header);
08f6c05feb1d~:arch/mips/net/bpf_jit_comp.c:
bpf_jit_binary_lock_ro(header);
08f6c05feb1d~:arch/parisc/net/bpf_jit_core.c:
bpf_jit_binary_lock_ro(jit_data->header);
08f6c05feb1d~:arch/s390/net/bpf_jit_comp.c:
bpf_jit_binary_lock_ro(header);
08f6c05feb1d~:arch/sparc/net/bpf_jit_comp_64.c:
bpf_jit_binary_lock_ro(header);
08f6c05feb1d~:arch/x86/net/bpf_jit_comp32.c:
bpf_jit_binary_lock_ro(header);
08f6c05feb1d~:include/linux/filter.h:static inline void
bpf_jit_binary_lock_ro(struct bpf_binary_header *hdr)

Therefore, commit 08f6c05feb1d should have included a backport for arm64.

So yes, I agree with Greg, the correct fix should be to backport to
ARM64 the changes done on other architectures in order to properly
handle return of set_memory_rox() in bpf_jit_binary_lock_ro().

Christophe