Skip to contentSkip to conversation

powerpc: fix the allyesconfig build

PATCH In review Stephen Rothwell 4 participants 8 messages Nov 28, 2020 – Nov 30, 2020

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 powerpc: fix the allyesconfig build

There are 2 drivers that have arrays of packed structures that contain pointers that end up at unaligned offsets. These produce warnings in the PowerPC allyesconfig build like this:

WARNING: 148 bad relocations
c00000000e56510b R_PPC64_UADDR64 .rodata+0x0000000001c72378
c00000000e565126 R_PPC64_UADDR64 .rodata+0x0000000001c723c0

They are not drivers that are used on PowerPC (I assume), so mark them to not be built on PPC64 when CONFIG_RELOCATABLE is enabled.

Cc Geert Uytterhoeven <[email protected]>
Cc Michael Turquette <[email protected]>
Cc Stephen Boyd <[email protected]>
Cc Yisen Zhuang <[email protected]>
Cc Salil Mehta <[email protected]>
Cc David S. Miller <[email protected]>
Cc Jakub Kicinski <[email protected]>
Cc Nicholas Piggin <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Joel Stanley <[email protected]>
---
 drivers/clk/renesas/Kconfig            | 4 ++++
 drivers/net/ethernet/hisilicon/Kconfig | 4 ++++
 2 files changed, 8 insertions(+)

It might be easiest to put this through the PowerPC (fixes?) tree, if people woulf just sned ACKs, please?

drivers/clk/renesas/Kconfig +4 -0
index 18915d668a30..53f24a0cad7a 100644
@@ -151,6 +151,10 @@ config CLK_R8A779A0
 	select CLK_RENESAS_CPG_MSSR
 
 config CLK_R9A06G032
+	# PPC64 RELOCATABLE kernels cannot handle relocations to
+	# unaligned locations that are produced by the array of
+	# packed structures in this driver.
+	depends on !(PPC64 && RELOCATABLE)
 	bool "Renesas R9A06G032 clock driver"
 	help
 	  This is a driver for R9A06G032 clocks
drivers/net/ethernet/hisilicon/Kconfig +4 -0
index 44f9279cdde1..bf47e504b365 100644
@@ -102,6 +102,10 @@ config HNS3_HCLGE
 	tristate "Hisilicon HNS3 HCLGE Acceleration Engine & Compatibility Layer Support"
 	default m
 	depends on PCI_MSI
+	# PPC64 RELOCATABLE kernels cannot handle relocations to
+	# unaligned locations that are produced by the array of
+	# packed structures in this driver.
+	depends on !(PPC64 && RELOCATABLE)
 	help
 	  This selects the HNS3_HCLGE network acceleration engine & its hardware
 	  compatibility layer. The engine would be used in Hisilicon hip08 family of
Signature · 5 lines
2.29.2

-- 
Cheers,
Stephen Rothwell
maintainer
#1 lore raw
Quoted 10 lines · On Sat, 28 Nov 2020 12:28:19 +1100 Stephen Rothwell wrote:

There are 2 drivers that have arrays of packed structures that contain pointers that end up at unaligned offsets. These produce warnings in the PowerPC allyesconfig build like this:

WARNING: 148 bad relocations
c00000000e56510b R_PPC64_UADDR64 .rodata+0x0000000001c72378
c00000000e565126 R_PPC64_UADDR64 .rodata+0x0000000001c723c0

They are not drivers that are used on PowerPC (I assume), so mark them to not be built on PPC64 when CONFIG_RELOCATABLE is enabled.

😳😳

What's the offending structure in hisilicon? I'd rather have a look packing structs with pointers in 'em sounds questionable.

I only see these two:

$ git grep packed drivers/net/ethernet/hisilicon/
drivers/net/ethernet/hisilicon/hns/hnae.h:struct __packed hnae_desc {
drivers/net/ethernet/hisilicon/hns3/hns3_enet.h:struct __packed hns3_desc {
#2 lore raw
Quoted 14 lines · Jakub Kicinski
Quoted 10 lines · On Sat, 28 Nov 2020 12:28:19 +1100 Stephen Rothwell wrote:

There are 2 drivers that have arrays of packed structures that contain pointers that end up at unaligned offsets. These produce warnings in the PowerPC allyesconfig build like this:

WARNING: 148 bad relocations
c00000000e56510b R_PPC64_UADDR64 .rodata+0x0000000001c72378
c00000000e565126 R_PPC64_UADDR64 .rodata+0x0000000001c723c0

They are not drivers that are used on PowerPC (I assume), so mark them to not be built on PPC64 when CONFIG_RELOCATABLE is enabled.

😳😳

Quoted 8 lines

What's the offending structure in hisilicon? I'd rather have a look packing structs with pointers in 'em sounds questionable.

I only see these two:

$ git grep packed drivers/net/ethernet/hisilicon/
drivers/net/ethernet/hisilicon/hns/hnae.h:struct __packed hnae_desc {
drivers/net/ethernet/hisilicon/hns3/hns3_enet.h:struct __packed hns3_desc {

I assmue "struct __packed hnae_desc" is the offending structure, because flag_ipoffset field is defined as __le32 and is not 32 bit aligned.

struct __packed hnae_desc {
	__le64 addr;							//0
	union {
		struct {						//64
			union {
				__le16 asid_bufnum_pid;
				__le16 asid;
			};
			__le16 send_size;				//92
			union {
				__le32 flag_ipoffset;			//*108*
				struct {
					__u8 bn_pid;
					__u8 ra_ri_cs_fe_vld;
					__u8 ip_offset;
					__u8 tse_vlan_snap_v6_sctp_nth;
				};
			};
			__le16 mss;
			__u8 l4_len;
			__u8 reserved1;
			__le16 paylen;
			__u8 vmid;
			__u8 qid;
			__le32 reserved2[2];
		} tx;

		struct {
			__le32 ipoff_bnum_pid_flag;
			__le16 pkt_len;
			__le16 size;
			union {
				__le32 vlan_pri_asid;
				struct {
					__le16 asid;
					__le16 vlan_cfi_pri;
				};
			};
			__le32 rss_hash;
			__le32 reserved_1[2];
		} rx;
	};
};
Quoted 2 lines

.

#3 lore raw
In reply to Jakub Kicinski

Hi Jakub,

Quoted 9 lines · On Fri, 27 Nov 2020 17:56:42 -0800 Jakub Kicinski <[email protected]> wrote:

What's the offending structure in hisilicon? I'd rather have a look packing structs with pointers in 'em sounds questionable.

I only see these two:

$ git grep packed drivers/net/ethernet/hisilicon/
drivers/net/ethernet/hisilicon/hns/hnae.h:struct __packed hnae_desc {
drivers/net/ethernet/hisilicon/hns3/hns3_enet.h:struct __packed hns3_desc {

struct hclge_dbg_reg_type_info which is 28 bytes long due to the included struct struct hclge_dbg_reg_common_msg (which is 12 bytes long). They are surrounded by #pragma pack(1)/pack().

This forces the 2 pointers in each second array element of hclge_dbg_reg_info[] to be 4 byte aligned (where pointers are 8 bytes long on PPC64).

Signature · 2 lines
Cheers,
Stephen Rothwell
maintainer
#4 lore raw
Quoted 11 lines · On Sat, 28 Nov 2020 16:20:54 +1100 Stephen Rothwell wrote:
Quoted 9 lines · On Fri, 27 Nov 2020 17:56:42 -0800 Jakub Kicinski <[email protected]> wrote:

What's the offending structure in hisilicon? I'd rather have a look packing structs with pointers in 'em sounds questionable.

I only see these two:

$ git grep packed drivers/net/ethernet/hisilicon/
drivers/net/ethernet/hisilicon/hns/hnae.h:struct __packed hnae_desc {
drivers/net/ethernet/hisilicon/hns3/hns3_enet.h:struct __packed hns3_desc {  
Quoted 7 lines

struct hclge_dbg_reg_type_info which is 28 bytes long due to the included struct struct hclge_dbg_reg_common_msg (which is 12 bytes long). They are surrounded by #pragma pack(1)/pack().

This forces the 2 pointers in each second array element of hclge_dbg_reg_info[] to be 4 byte aligned (where pointers are 8 bytes long on PPC64).

Ah! Thanks, I don't see a reason for these to be packed. Looks like an accident, there is no reason to pack anything past struct hclge_dbg_reg_common_msg AFAICT.

Huawei folks, would you mind sending a fix if the analysis is correct?

#5 lore raw
Quoted 20 lines · Jakub Kicinski
Quoted 11 lines · On Sat, 28 Nov 2020 16:20:54 +1100 Stephen Rothwell wrote:
Quoted 9 lines · On Fri, 27 Nov 2020 17:56:42 -0800 Jakub Kicinski <[email protected]> wrote:

What's the offending structure in hisilicon? I'd rather have a look packing structs with pointers in 'em sounds questionable.

I only see these two:

$ git grep packed drivers/net/ethernet/hisilicon/
drivers/net/ethernet/hisilicon/hns/hnae.h:struct __packed hnae_desc {
drivers/net/ethernet/hisilicon/hns3/hns3_enet.h:struct __packed hns3_desc {  
Quoted 7 lines

struct hclge_dbg_reg_type_info which is 28 bytes long due to the included struct struct hclge_dbg_reg_common_msg (which is 12 bytes long). They are surrounded by #pragma pack(1)/pack().

This forces the 2 pointers in each second array element of hclge_dbg_reg_info[] to be 4 byte aligned (where pointers are 8 bytes long on PPC64).

Quoted 5 lines

Ah! Thanks, I don't see a reason for these to be packed. Looks like an accident, there is no reason to pack anything past struct hclge_dbg_reg_common_msg AFAICT.

Huawei folks, would you mind sending a fix if the analysis is correct?

Yes, will send a patch to fix that. Thanks for the analysis.

Quoted 2 lines

.

#6 lore raw
In reply to Stephen Rothwell

Hi Stephen,

Quoted 11 lines · Stephen Rothwell

There are 2 drivers that have arrays of packed structures that contain pointers that end up at unaligned offsets. These produce warnings in the PowerPC allyesconfig build like this:

WARNING: 148 bad relocations
c00000000e56510b R_PPC64_UADDR64 .rodata+0x0000000001c72378
c00000000e565126 R_PPC64_UADDR64 .rodata+0x0000000001c723c0

They are not drivers that are used on PowerPC (I assume), so mark them to not be built on PPC64 when CONFIG_RELOCATABLE is enabled.

Quoted 11 lines
Cc Geert Uytterhoeven <[email protected]>
Cc Michael Turquette <[email protected]>
Cc Stephen Boyd <[email protected]>
Cc Yisen Zhuang <[email protected]>
Cc Salil Mehta <[email protected]>
Cc David S. Miller <[email protected]>
Cc Jakub Kicinski <[email protected]>
Cc Nicholas Piggin <[email protected]>
Cc Daniel Axtens <[email protected]>
Cc Joel Stanley <[email protected]>

Thanks for your patch!

drivers/clk/renesas/Kconfig config CLK_R8A779A0 · line 160

Quoted 13 lines
drivers/clk/renesas/Kconfig +4 -0
@@ -151,6 +151,10 @@ config CLK_R8A779A0
        select CLK_RENESAS_CPG_MSSR
 
 config CLK_R9A06G032
+       # PPC64 RELOCATABLE kernels cannot handle relocations to
+       # unaligned locations that are produced by the array of
+       # packed structures in this driver.
+       depends on !(PPC64 && RELOCATABLE)
        bool "Renesas R9A06G032 clock driver"
        help
          This is a driver for R9A06G032 clocks

I prefer to fix this in the driver instead. The space saving by packing the structure is minimal.
I've sent a patch
https://lore.kernel.org/r/[email protected]
(when lore is back)

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#7 lore raw

Hi Geert,

Quoted 2 lines · On Mon, 30 Nov 2020 09:58:23 +0100 Geert Uytterhoeven <[email protected]>…

Thanks for your patch!

No worries, it has been a small irritant to me for quite a while.

Quoted 5 lines

I prefer to fix this in the driver instead. The space saving by packing the structure is minimal.
I've sent a patch
https://lore.kernel.org/r/[email protected]
(when lore is back)

Absolutely, thanks.

Signature · 2 lines
Cheers,
Stephen Rothwell