From: Wentao Guan <guanwentao@uniontech.com>
To: chenhuacai@kernel.org
Cc: wuqianhai@loongson.cn, kernel@xen0n.name,
jiaxun.yang@flygoat.com, loongarch@lists.linux.dev,
linux-kernel@vger.kernel.org,
Wentao Guan <guanwentao@uniontech.com>,
stable@vger.kernel.org
Subject: [PATCH v4] LoongArch: Fix potential ade in loongson_gpu_fixup_dma_hang()
Date: Tue, 28 Apr 2026 19:01:55 +0800 [thread overview]
Message-ID: <20260428110155.754875-1-guanwentao@uniontech.com> (raw)
The switch case in loongson_gpu_fixup_dma_hang() may not DC2 or DC3,
and readl(crtc_reg) will access with random address,
because device is from base+PCI_DEVICE_ID, base is from pdev->devfn+1,
it is wrong when my platform inserts a gpu:
lspci -tv
-[0000:00]-+-00.0 Loongson Technology LLC Hyper Transport Bridge Controller
...
+-06.0 Loongson Technology LLC LG100 GPU
+-06.2 Loongson Technology LLC Device 7a37
...
Add a default switch case to fix it.
It not a issue in v7.1-rc1, but stil cause the problem in v6.6.136.
In v7.1-rc1:
[ 0.817545] pci 0000:00:06.0: Failed to ioremap()
[ 0.822215] pci 0000:00:06.0: [0014:7a25] type 00 class 0x040000 conventional PCI endpoint
[ 0.830434] pci 0000:00:06.0: BAR 0 [mem 0xe8025162000-0xe80251620ff 64bit]
[ 0.837350] pci 0000:00:06.0: BAR 2 [mem 0xe8010000000-0xe801fffffff 64bit]
[ 0.844267] pci 0000:00:06.0: BAR 4 [mem 0xe8025120000-0xe802512ffff 64bit]
[ 0.851214] pci 0000:00:06.2: [0014:7a37] type 00 class 0x040300 conventional PCI endpoint
[ 0.859433] pci 0000:00:06.2: BAR 0 [mem 0xe8025110000-0xe802511ffff 64bit]
In v6.6.136 before:
[ 0.807099] Kernel ade access[#1]:
[ 0.810472] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 6.6.136-loong64-desktop-hwe+ #4
[ 0.818252] Hardware name: Loongson Loongson-3A6000-HV-7A2000-1w-V0.1-EVB/Loongson-3A6000-HV-7A2000-1w-EVB-V1.21, BIOS Loongson-UDK2018-V4.0.05756-prestab
[ 0.831992] pc 90000000017e5534 ra 90000000017e54c0 tp 90000001002f8000 sp 90000001002fb6c0
[ 0.840289] a0 80000efe00003100 a1 0000000000003100 a2 0000000000000000 a3 0000000000000002
[ 0.848585] a4 90000001002fb6b4 a5 900000087cdb58fd a6 90000000027af000 a7 0000000000000001
[ 0.856882] t0 00000000000085b9 t1 000000000000ffff t2 0000000000000000 t3 0000000000000000
[ 0.865179] t4 fffffffffffffffd t5 00000000fffb6d9c t6 0000000000083b00 t7 00000000000070c0
[ 0.873475] t8 900000087cdb4d94 u0 900000087cdb58fd s9 90000001002fb826 s0 90000000031c12c8
[ 0.881771] s1 7fffffffffffff00 s2 90000000031c12d0 s3 0000000000002710 s4 0000000000000000
[ 0.890067] s5 0000000000000000 s6 9000000100053000 s7 7fffffffffffff00 s8 90000000030d4000
[ 0.898364] ra: 90000000017e54c0 loongson_gpu_fixup_dma_hang+0x40/0x210
[ 0.905195] ERA: 90000000017e5534 loongson_gpu_fixup_dma_hang+0xb4/0x210
[ 0.912023] CRMD: 000000b0 (PLV0 -IE -DA +PG DACF=CC DACM=CC -WE)
[ 0.918165] PRMD: 00000004 (PPLV0 +PIE -PWE)
[ 0.922489] EUEN: 00000000 (-FPE -SXE -ASXE -BTE)
[ 0.927246] ECFG: 00071c1d (LIE=0,2-4,10-12 VS=7)
[ 0.932002] ESTAT: 00480000 [ADEM] (IS= ECode=8 EsubCode=1)
[ 0.937535] BADV: 7fffffffffffff00
[ 0.940992] PRID: 0014d000 (Loongson-64bit, Loongson-3A6000-HV)
[ 0.946956] Modules linked in:
[ 0.949982] Process swapper/0 (pid: 1, threadinfo=(____ptrval____), task=(____ptrval____))
[ 0.958193] Stack : 0000000000000006 90000001002fb778 90000001002fb704 0000000000000007
[ 0.966147] 0000000016a65700 90000000017e5690 000000000000ffff ffffffffffffffff
[ 0.974100] 900000000209f7c0 9000000100053000 900000000209f7a8 9000000000eebc08
[ 0.982053] 0000000000000000 0000000000000000 0000000000000006 90000001002fb778
[ 0.990006] 90000001000530b8 90000000027af000 0000000000000000 9000000100054000
[ 0.997959] 9000000100053000 9000000000ebb70c 9000000100004c00 9000000004000001
[ 1.005913] 90000001002fb7e4 bae765461f31cb12 0000000000000000 0000000000000000
[ 1.013866] 0000000000000006 90000000027af000 0000000000000030 90000000027af000
[ 1.021819] 900000087cd6f800 9000000100053000 0000000000000000 9000000000ebc560
[ 1.029772] 7a2500147cdaf720 bae765461f31cb12 0000000000000001 0000000000000030
[ 1.037725] ...
[ 1.040146] Call Trace:
[ 1.040148] [<90000000017e5534>] loongson_gpu_fixup_dma_hang+0xb4/0x210
[ 1.049138] [<9000000000eebc08>] pci_fixup_device+0x108/0x280
[ 1.054846] [<9000000000ebb70c>] pci_setup_device+0x24c/0x690
[ 1.060551] [<9000000000ebc560>] pci_scan_single_device+0xe0/0x140
[ 1.066688] [<9000000000ebc684>] pci_scan_slot+0xc4/0x280
[ 1.072048] [<9000000000ebdd00>] pci_scan_child_bus_extend+0x60/0x3f0
[ 1.078444] [<9000000000f5bc94>] acpi_pci_root_create+0x2b4/0x420
[ 1.084498] [<90000000017e5e74>] pci_acpi_scan_root+0x2d4/0x440
[ 1.090376] [<9000000000f5b02c>] acpi_pci_root_add+0x21c/0x3a0
[ 1.096168] [<9000000000f4ee54>] acpi_bus_attach+0x1a4/0x3c0
[ 1.101788] [<90000000010e200c>] device_for_each_child+0x6c/0xe0
[ 1.107755] [<9000000000f4bbf4>] acpi_dev_for_each_child+0x44/0x70
[ 1.113892] [<9000000000f4ef40>] acpi_bus_attach+0x290/0x3c0
[ 1.119511] [<90000000010e200c>] device_for_each_child+0x6c/0xe0
[ 1.125476] [<9000000000f4bbf4>] acpi_dev_for_each_child+0x44/0x70
[ 1.131612] [<9000000000f4ef40>] acpi_bus_attach+0x290/0x3c0
[ 1.137231] [<9000000000f5211c>] acpi_bus_scan+0x6c/0x280
[ 1.142591] [<900000000189c028>] acpi_scan_init+0x194/0x310
[ 1.148125] [<900000000189bc6c>] acpi_init+0xcc/0x140
[ 1.153139] [<9000000000220cdc>] do_one_initcall+0x4c/0x310
[ 1.158672] [<90000000018618fc>] kernel_init_freeable+0x258/0x2d4
[ 1.164726] [<900000000184326c>] kernel_init+0x28/0x13c
[ 1.169914] [<9000000000222008>] ret_from_kernel_thread+0xc/0xa4
[ 1.175878]
[ 1.177349] Code: 0015001b 02c022f9 0010efd8 <2400030c> 0040818c 38720005 40006380 240002ed 034401ad
[ 1.187034]
After:
[ 0.813002] pci 0000:00:06.0: [0014:7a25] type 00 class 0x040000
[ 0.818970] pci 0000:00:06.0: BAR 0 [mem 0xe8025162000-0xe80251620ff 64bit]
[ 0.825887] pci 0000:00:06.0: BAR 2 [mem 0xe8010000000-0xe801fffffff 64bit]
[ 0.832804] pci 0000:00:06.0: BAR 4 [mem 0xe8025120000-0xe802512ffff 64bit]
[ 0.839750] pci 0000:00:06.2: [0014:7a37] type 00 class 0x040300
[ 0.845718] pci 0000:00:06.2: BAR 0 [mem 0xe8025110000-0xe802511ffff 64bit]
Cc: stable@vger.kernel.org
Fixes: 95db0c9f526d ("LoongArch: Workaround LS2K/LS7A GPU DMA hang bug")
Link: https://gist.github.com/opsiff/ebf2dac51b4013d22462f2124c55f807
Link: https://gist.github.com/opsiff/a62f2a73db0492b3c49bf223a339b133
Signed-off-by: Wentao Guan <guanwentao@uniontech.com>
---
changelog v4:
rollback to v2, because need the regbase in switch-case.
changelog v3:
test in v7.1-rc1 and remove unused print for it can be read from lspci -tv.
changelog v2:
reformat commit msg and add a full dmesg log link to it.
---
---
arch/loongarch/pci/pci.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/arch/loongarch/pci/pci.c b/arch/loongarch/pci/pci.c
index d233ea2218fe0..f33c7ea1443d9 100644
--- a/arch/loongarch/pci/pci.c
+++ b/arch/loongarch/pci/pci.c
@@ -132,6 +132,9 @@ static void loongson_gpu_fixup_dma_hang(struct pci_dev *pdev, bool on)
crtc_reg = regbase;
crtc_offset = 0x400;
break;
+ default:
+ iounmap(regbase);
+ return;
}
for (i = 0; i < CRTC_NUM_MAX; i++, crtc_reg += crtc_offset) {
--
2.30.2
next reply other threads:[~2026-04-28 11:04 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-28 11:01 Wentao Guan [this message]
2026-04-29 14:25 ` [PATCH v4] LoongArch: Fix potential ade in loongson_gpu_fixup_dma_hang() Huacai Chen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260428110155.754875-1-guanwentao@uniontech.com \
--to=guanwentao@uniontech.com \
--cc=chenhuacai@kernel.org \
--cc=jiaxun.yang@flygoat.com \
--cc=kernel@xen0n.name \
--cc=linux-kernel@vger.kernel.org \
--cc=loongarch@lists.linux.dev \
--cc=stable@vger.kernel.org \
--cc=wuqianhai@loongson.cn \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox