linux-kernel - [PATCH] arm64: patching: fix handling of execmem addresses

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: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <20240521213813.703309-1-rppt@kernel.org>
Date: Wed, 22 May 2024 00:38:13 +0300
From: Mike Rapoport <rppt@...nel.org>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: Catalin Marinas <catalin.marinas@....com>,
	Klara Modin <klarasmodin@...il.com>,
	Mike Rapoport <rppt@...nel.org>,
	Will Deacon <will@...nel.org>,
	linux-arm-kernel@...ts.infradead.org,
	linux-mm@...ck.org,
	linux-kernel@...r.kernel.org
Subject: [PATCH] arm64: patching: fix handling of execmem addresses

From: Will Deacon <will@...nel.org>

Klara Modin reported warnings for a kernel configured with BPF_JIT but
without MODULES:

[   44.131296] Trying to vfree() bad address (000000004a17c299)
[   44.138024] WARNING: CPU: 1 PID: 193 at mm/vmalloc.c:3189 remove_vm_area (mm/vmalloc.c:3189 (discriminator 1))
[   44.146675] CPU: 1 PID: 193 Comm: kworker/1:2 Tainted: G      D W          6.9.0-01786-g2c9e5d4a0082 #25
[   44.158229] Hardware name: Raspberry Pi 3 Model B (DT)
[   44.164433] Workqueue: events bpf_prog_free_deferred
[   44.170492] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[   44.178601] pc : remove_vm_area (mm/vmalloc.c:3189 (discriminator 1))
[   44.183705] lr : remove_vm_area (mm/vmalloc.c:3189 (discriminator 1))
[   44.188772] sp : ffff800082a13c70
[   44.193112] x29: ffff800082a13c70 x28: 0000000000000000 x27: 0000000000000000
[   44.201384] x26: 0000000000000000 x25: ffff00003a44efa0 x24: 00000000d4202000
[   44.209658] x23: ffff800081223dd0 x22: ffff00003a198a40 x21: ffff8000814dd880
[   44.217924] x20: 00000000d4202000 x19: ffff8000814dd880 x18: 0000000000000006
[   44.226206] x17: 0000000000000000 x16: 0000000000000020 x15: 0000000000000002
[   44.234460] x14: ffff8000811a6370 x13: 0000000020000000 x12: 0000000000000000
[   44.242710] x11: ffff8000811a6370 x10: 0000000000000144 x9 : ffff8000811fe370
[   44.250959] x8 : 0000000000017fe8 x7 : 00000000fffff000 x6 : ffff8000811fe370
[   44.259206] x5 : 0000000000000000 x4 : 0000000000000000 x3 : 0000000000000000
[   44.267457] x2 : 0000000000000000 x1 : 0000000000000000 x0 : ffff000002203240
[   44.275703] Call trace:
[   44.279158] remove_vm_area (mm/vmalloc.c:3189 (discriminator 1))
[   44.283858] vfree (mm/vmalloc.c:3322)
[   44.287835] execmem_free (mm/execmem.c:70)
[   44.292347] bpf_jit_free_exec+0x10/0x1c
[   44.297283] bpf_prog_pack_free (kernel/bpf/core.c:1006)
[   44.302457] bpf_jit_binary_pack_free (kernel/bpf/core.c:1195)
[   44.307951] bpf_jit_free (include/linux/filter.h:1083 arch/arm64/net/bpf_jit_comp.c:2474)
[   44.312342] bpf_prog_free_deferred (kernel/bpf/core.c:2785)
[   44.317785] process_one_work (kernel/workqueue.c:3273)
[   44.322684] worker_thread (kernel/workqueue.c:3342 (discriminator 2) kernel/workqueue.c:3429 (discriminator 2))
[   44.327292] kthread (kernel/kthread.c:388)
[   44.331342] ret_from_fork (arch/arm64/kernel/entry.S:861)

The problem is because bpf_arch_text_copy() silently fails
to write to the read-only area as a result of patch_map() faulting and
the resulting -EFAULT being chucked away.

Update patch_map() to use CONFIG_EXECMEM instead of
CONFIG_STRICT_MODULE_RWX to check for vmalloc addresses.

Fixes: 2c9e5d4a0082 ("bpf: remove CONFIG_BPF_JIT dependency on CONFIG_MODULES of")
Reported-by: Klara Modin <klarasmodin@...il.com>
Closes: https://lore.kernel.org/all/7983fbbf-0127-457c-9394-8d6e4299c685@gmail.com
Signed-off-by: Will Deacon <will@...nel.org>
Tested-by: Klara Modin <klarasmodin@...il.com>
Signed-off-by: Mike Rapoport (IBM) <rppt@...nel.org>
---

Andrew, if you can pick this up before -rc1 it won't need to wait for
arm64 tree rebase

 arch/arm64/kernel/patching.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/patching.c b/arch/arm64/kernel/patching.c
index 255534930368..945df74005c7 100644
--- a/arch/arm64/kernel/patching.c
+++ b/arch/arm64/kernel/patching.c
@@ -36,7 +36,7 @@ static void __kprobes *patch_map(void *addr, int fixmap)
 
 	if (image)
 		page = phys_to_page(__pa_symbol(addr));
-	else if (IS_ENABLED(CONFIG_STRICT_MODULE_RWX))
+	else if (IS_ENABLED(CONFIG_EXECMEM))
 		page = vmalloc_to_page(addr);
 	else
 		return addr;

base-commit: 8f6a15f095a63a83b096d9b29aaff4f0fbe6f6e6
-- 
2.43.0