Re: [f2fs-dev] Kernel BUG when writing to f2fs drive, PowerPC, SD card, USB3

[Huajun Li <huajun.li.lee@gmail.com>]

Hi,
Is it possible caused by bitmap ? you know, bitmaps are unsigned
variable, while f2fs_{clear, set, test}_bit() parameter is signed
variable.
So, could you please try following patch, and update the same issue in
mkfs.f2fs tool.

diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 3e7cb33..28d31f1 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -209,7 +209,7 @@ struct f2fs_nm_info {
     struct mutex build_lock;    /* lock for build free nids */

     /* for checkpoint */
-    char *nat_bitmap;        /* NAT bitmap pointer */
+    unsigned char *nat_bitmap;    /* NAT bitmap pointer */
     int bitmap_size;        /* bitmap size */
 };

@@ -820,7 +820,7 @@ static inline block_t datablock_addr(struct page *node_page,
     return le32_to_cpu(addr_array[offset]);
 }

-static inline int f2fs_test_bit(unsigned int nr, char *addr)
+static inline int f2fs_test_bit(unsigned int nr, unsigned char *addr)
 {
     int mask;

@@ -829,7 +829,7 @@ static inline int f2fs_test_bit(unsigned int nr, char *addr)
     return mask & *addr;
 }

-static inline int f2fs_set_bit(unsigned int nr, char *addr)
+static inline int f2fs_set_bit(unsigned int nr, unsigned char *addr)
 {
     int mask;
     int ret;
@@ -841,7 +841,7 @@ static inline int f2fs_set_bit(unsigned int nr, char *addr)
     return ret;
 }

-static inline int f2fs_clear_bit(unsigned int nr, char *addr)
+static inline int f2fs_clear_bit(unsigned int nr, unsigned char *addr)
 {
     int mask;
     int ret;
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index b15debc..b191449 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -1402,7 +1402,7 @@ static int build_sit_info(struct f2fs_sb_info *sbi)
     struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
     struct sit_info *sit_i;
     unsigned int sit_segs, start;
-    char *src_bitmap, *dst_bitmap;
+    unsigned char *src_bitmap, *dst_bitmap;
     unsigned int bitmap_size;

     /* allocate memory for SIT information */
diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
index 062424a..a96f7f4 100644
--- a/fs/f2fs/segment.h
+++ b/fs/f2fs/segment.h
@@ -175,7 +175,7 @@ struct sit_info {
     block_t sit_base_addr;        /* start block address of SIT area */
     block_t sit_blocks;        /* # of blocks used by SIT area */
     block_t written_valid_blocks;    /* # of valid blocks in main area */
-    char *sit_bitmap;        /* SIT bitmap pointer */
+    unsigned char *sit_bitmap;    /* SIT bitmap pointer */
     unsigned int bitmap_size;    /* SIT bitmap size */

     unsigned long *dirty_sentries_bitmap;    /* bitmap for dirty sentries */

On Sun, Jun 16, 2013 at 7:51 PM, Oded Gabbay <ogabbay@advaoptical.com> wrote:
> Hi,
>
> I'm working on a custom board with a PowerPC processor (Freescale P2020).
> On the board there is an SD card, which is connected to a USB3 chip (from
> TI), which is connected to the PCI-e controller of the CPU.
> I'm running with Linux kernel 3.9.6, with our custom rootFS.
>
> I formatted an SD card using the mkfs.f2fs utility (after fixing some
> Big-endian issues - sent a patch a few days ago).
> I then mounted the SD card, using "mount -o
> noatime,nodiratime,rw,nosuid,nodev,relatime,active_logs=6,uhelper=udisks2,background_gc_off
> /dev/sda /mnt/sd1"
> Then, I started a small user-space test application which opens a file on
> the mount folder and starts to do "fwrite" into the file.
> After 2-3 seconds, the kernel gives me a BUG and the system restarts.
> When the system is up and I try to re-mount the SD card, I get the following
> error message:
>
> F2FS-fs (sda): Failed to get valid F2FS checkpoint
> mount: you must specify the filesystem type
>
> Only way is to re-format the card using mkfs.f2fs
>
> I took the f2fs patch that Jaegeuk Kim sent to Linus for 3.10 (here -
> https://lkml.org/lkml/2013/5/8/122) and applied it cleanly to 3.9.6
> I repeated the procedure but got the same result.
>
> The BUG is from this line, from segment.c:
>         if (!f2fs_clear_bit(offset, se->cur_valid_map))
>             BUG();
>
> Additional information I can give is
>
> 1. I tried using F2FS in ArchLinux, kernel 3.9.5, on an x86 machine, with
> the same SD card and the same USB3-to-PCIe chip and it worked flawlessly
> there.
> 2. I can work with other FS on the SD card on our custom board, such as
> Ext3, Ext4 and vfat, so this is not a H/W issue.
>
> Could you please try to help me pinpoint/debug the problem ?
>
> Here is the complete kernel BUG print:
>
> kernel BUG at .../linux-3.9.6-adva/fs/f2fs/segment.c:214!
> Oops: Exception in kernel mode, sig: 5 [#1]
> PREEMPT SMP NR_CPUS=2 P2020 FSP150
> Modules linked in: mdio(O) hardware_version(PO) clipresent(PO) monotonic(O)
> restartcause(PO) panic_buffer(O)
> NIP: c026a7e0 LR: c026a660 CTR: 00000000
> REGS: ee761a60 TRAP: 0700   Tainted: P           O
> (3.9.6-dev_ogabbay-109482*)
> MSR: 00029000 <CE,EE,ME>  CR: 24a52588  XER: 20000000
> TASK = efb444c0[1755] 'flush-8:0' THREAD: ee760000 CPU: 1
> GPR00: 00000000 ee761b10 efb444c0 0000004c 00000000 00000000 01dc4900
> eb0fa700
> GPR08: 00000000 eb24cb00 00000040 00000040 00000038 00000000 ee761c64
> 00000000
> GPR16: c0aeea80 00080000 00000000 00000000 0000ed31 eb0fa700 00000000
> eb0fa734
> GPR24: eb0fa700 00000080 f2030620 ffffffff ffffffc8 0000ed31 ffffffff
> c55a1000
> NIP [c026a7e0] update_sit_entry+0x240/0x248
> LR [c026a660] update_sit_entry+0xc0/0x248
> Call Trace:
> [ee761b10] [c55a1000] 0xc55a1000 (unreliable)
> [ee761b40] [c026d1f4] do_write_page+0x198/0x660
> [ee761b80] [c026d84c] write_data_page+0xa4/0xb8
> [ee761bc0] [c0265118] do_write_data_page+0x1e8/0x20c
> [ee761c20] [c02653dc] f2fs_write_data_page+0x2a0/0x2c0
> [ee761c40] [c0263ad8] __f2fs_writepage+0x24/0x80
> [ee761c50] [c00b05dc] write_cache_pages+0x1d0/0x35c
> [ee761d00] [c0263cf4] f2fs_write_data_pages+0xf4/0xfc
> [ee761d30] [c00b1d3c] do_writepages+0x30/0x64
> [ee761d40] [c0103fbc] __writeback_single_inode+0x34/0x10c
> [ee761d60] [c0104ef8] writeback_sb_inodes+0x204/0x370
> [ee761dd0] [c01050f4] __writeback_inodes_wb+0x90/0xd4
> [ee761e00] [c01054cc] wb_writeback+0x204/0x20c
> [ee761e50] [c0105844] wb_do_writeback+0x144/0x20c
> [ee761eb0] [c0105980] bdi_writeback_thread+0x74/0x144
> [ee761ee0] [c0059dc4] kthread+0xa8/0xac
> [ee761f40] [c000f014] ret_from_kernel_thread+0x64/0x6c
> Instruction dump:
> 4bffff2c 813a0004 5720e8fe 7f39c8f8 39400001 5739077e 7d6900ae 7d59c830
> 7d6ac878 7d68c839 7d4901ae 4082fef0 <0fe00000> 0fe00000 9421ffe0 7c0802a6
>
> --
> Best regards,
> Oded Gabbay
> Principal Engineer Advanced Packet Technologies
> ADVA Optical Networking Israel Ltd.
> P.O. Box 2552
> 2 Hatidhar St.
> Raanana 4366504, Israel
> Tel: +(972)-9-7750130
> Fax: +(972)-9-7462092
> Mobile: +(972)-54-6543998
> E-mail: ogabbay@advaoptical.com
>
> www.advaoptical.com
> Let's ADVANCE
>
> ADVA Optical Networking SE is a European stock corporation (\"Societas
> Europaea\") with registered offices at Maerzenquelle 1-3, D-98617 Meiningen,
> Germany * CEO: Brian L. Protiva, Chief Officers: Dr. Christoph Glingener,
> Christian Unterberger, Jaswir Singh * Chairman of the Supervisory Board:
> Anthony Maher * AG Jena HRB 508155 * VAT No. DE 175 446 349
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by Windows:
>
> Build for Windows Store.
>
> http://p.sf.net/sfu/windows-dev2dev
> _______________________________________________
> Linux-f2fs-devel mailing list
> Linux-f2fs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
>

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev