July 2011 - kernelnewbies@kernelnewbies.org - Kernelnewbies Mailing Lists

flush_cache_all API
by naveen yadav July 15, 2011

July 15, 2011

Hi All, I am using Multi-core ARM machine. If I use flush_cache_all() API, will it flush only current CPU cache ?, or flush all cache's of all core's. Thanks

4 4

Puzzled about MDIO issue
by bill July 15, 2011

July 15, 2011

Hi, All I've been reading MDIO code, tiny though but it still puzzled me for its usage. MDIO is an hardware interface for the MAC to set/get PHY configuration/status. Linux manipulates MDIO through: struct net_device ->do_ioctl -> mdio_mii_ioctl -> mdio_read/mdio_write Question: A: Turn an NIC from cold state to running state, is this ioctl must be used to configure the PHY? If so , how? B: what's the usage of this net device ioctl? I found an user-space package mii-tool create a socket first and then use the socket handle for ioctl to query lots of PHY information. Is there any other typical usage of it? thanks bill

2 2

SATA/SAS link speed negotiation
by Peter Hamilton July 14, 2011

July 14, 2011

I'm trying to find out what the negotiated speed is on a SATA port. dmesg reports the negotiated speed, but I'm hoping to find it somewhere in /sys or /proc (or some other tool). Scraping dmesg is a messy process as I'm starting with a block device. I would have to trace the block device to the scsi address and translate that to an ata address. Here's the scsi and block device info from dmesg: scsi 7:0:0:0: Direct-Access ATA WDC WD1600AAJS-2 01.0 PQ: 0 ANSI: 5 sd 7:0:0:0: [sdd] 312581808 512-byte logical blocks: (160 GB/149 GiB) sd 7:0:0:0: [sdd] Write Protect is off sd 7:0:0:0: [sdd] Mode Sense: 00 3a 00 00 sd 7:0:0:0: [sdd] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA sd 7:0:0:0: [sdd] Attached SCSI disk sd 7:0:0:0: Attached scsi generic sg3 type 0 Here's the corresponding ata info from dmesg: ata7: SATA link up 3.0 Gbps (SStatus 123 SControl 370) ata7.00: ATA-8: WDC WD1600AAJS-22L7A0, 01.03E01, max UDMA/133 ata7.00: 312581808 sectors, multi 0: LBA48 NCQ (depth 31/32), AA ata7.00: configured for UDMA/133 Finding the SATA link speed in dmesg would be much cleaner than scraping dmesg. Does anyone know where I might find that? Also, for a SAS HBA, the link speed is not reported to dmesg. Does anyone know how to find the link speed for a SAS drive? Thanks, Peter

2 2

driver ring buffer overrun and printf usage
by Amit Nagal July 14, 2011

July 14, 2011

Hi , In my embedded linux-arm target , in a userspace thread , i am reading hid events over usb-hiddev interface (/dev/hiddev) . basically what i do is this : read_thread(){ while(1){ ret = read( "/dev/hiddev" , user_buffer , bytes_to_read); printf("bytes received =%d" , ret ) ; // (printf prints debug message to serial console ); } } now whenever i use the printf statement in read thread , kernel space ring buffer in usb-hid driver is overrun frequently , which results in loss of data send by usb device . if i remove printf in read thread , no overruns are reported in usb-hid driver . that means if printf is used , application read_thread is not able to read data in time which leads to driver ring buffer overrun . does this behaviour of printf points to some problem in embedded linux's scheduler ( printf prints data to debug serial port , and consecutive read() calls in read_thread() are scheduled lately due to that ) ? do the actual implementation of printf involves any kind of sleeps ? kindly share necessary insight wrt above printf usage and driver ring buffer overrun . regards Amit Nagal

3 2

Why is the clear_bit() a specical bitop?
by Parmenides July 14, 2011

July 14, 2011

Hi, When reading asm/bitops.h, I have some questions about bit clear operations. p.s. the link: http://lxr.linux.no/linux+v2.6.34/arch/x86/include/asm/bitops.h The clear_bit() is defined as follows: /** * clear_bit - Clears a bit in memory * @nr: Bit to clear * @addr: Address to start counting from * * clear_bit() is atomic and may not be reordered. However, it does * not contain a memory barrier, so if it is used for locking purposes, * you should call smp_mb__before_clear_bit() and/or smp_mb__after_clear_bit() * in order to ensure changes are visible on other processors. */ static __always_inline void clear_bit(int nr, volatile unsigned long *addr) { if (IS_IMMEDIATE(nr)) { asm volatile(LOCK_PREFIX "andb %1,%0" : CONST_MASK_ADDR(nr, addr) : "iq" ((u8)~CONST_MASK(nr))); } else { asm volatile(LOCK_PREFIX "btr %1,%0" : BITOP_ADDR(addr) : "Ir" (nr)); } } 1. Two its parallel funcitons, namely set_bit() and change_bit() contain their memory barriers, while the clear_bit() does not has one. What does make it deserver such a special consideration? 2. How is it used for locking purpose? Is there any example? The clear_bit_unlock() is defined as follows: /* * clear_bit_unlock - Clears a bit in memory * @nr: Bit to clear * @addr: Address to start counting from * * clear_bit() is atomic and implies release semantics before the memory * operation. It can be used for an unlock. */ static inline void clear_bit_unlock(unsigned nr, volatile unsigned long *addr) { barrier(); clear_bit(nr, addr); } 3. "clear_bit() is atomic and implies ***release semantics*** before the memory operation" What's the meaning of "release semantics"? 4. Why does the barrier() precede the clear_bit()? AFAIK, barrier() will cause memory values reload into registers. Now that the clear_bit() modifies the memory, why is there no a barrier() after the clear_bit()?

1 0

how to apply patches from git ?
by RKK July 14, 2011

July 14, 2011

Hi all, Sorry if this is a basic question. How do i take patches from e-mail? for example git send mail sends patches as e-mail if someone wants to apply that patch to his branch then how do i get that in . patch format and then use patch -p1 *.patch or is there something im missing here? thanks. -- Warm Regards, Ravi .

8 12

Device mapper header file error
by Adil Mujeeb July 14, 2011

July 14, 2011

Hi List,**** ** ** I am trying to compile a kernel module which uses the device mapper header files (dm.h and dm-bio-list.h). I checked in the stock kernel source that these files are present under drivers/md directory.**** When I try to build the module, it gives error for these header files as missing. I checked the Makefile and found that it includes the path (-I$(TOPDIR)/ drivers/md)**** ** ** But when I check the content of this directory on my machine it doesn’t include the header files:**** [root@localhost redhat]# ls /usr/src/kernels/2.6.18-8.el5-i686/drivers/md*** * Kconfig Makefile raid6test**** [root@localhost redhat]#**** ** ** While the other header files referred by the kernel module under include/linux directory exists.**** ** ** My machine detail:**** [adil@localhost ~]$ uname -a**** Linux localhost.localdomain 2.6.18-8.el5 #1 SMP Fri Jan 26 14:15:21 EST 2007 i686 i686 i386 GNU/Linux**** [adil@localhost ~]$**** ** ** Thanks in advance.**** ** Regards, Adil Mujeeb

2 1

Puzzled about MDIO issue
by bill July 14, 2011

July 14, 2011

Hi, All I've been reading MDIO code, tiny though but it still puzzled me for its usage. MDIO is an hardware interface for the MAC to set/get PHY configuration/status. Linux manipulates MDIO through: struct net_device ->do_ioctl -> mdio_mii_ioctl -> mdio_read/mdio_write Question: A: Turn an NIC from cold state to running state, is this ioctl must be used to configure the PHY? If so , how? B: what's the usage of this net device ioctl? I found an user-space package mii-tool create a socket first and then use the socket handle for ioctl to query lots of PHY information. Is there any other typical usage of it? thanks bill

1 0

query regarding inode pages
by shubham sharma July 14, 2011

July 14, 2011

I am trying to write a memory based file system. The file system is intended to create files/directories and write their contents only on pages. For this I have used the function grab_cache_page() function to get a new locked page in case the page does not exists in the radix tree of the inode. As the filesystem is memory based and all the data exists only on the memory, so I don't release the lock on the page as I fear that the pdflush daemon can swap out the pages on which I have written data and I may never see that data again. I unlock all the pages of all the inodes of the file system in the kill_sb function once the file system is being unmounted. But the problem I am facing is that if I open a file in which I have already written something (and its pages are locked), the open system call in turn calls the __lock_page() function which waits for the pages belonging to the inode to get unlocked. Hence the system call stalls indefinitely. I want to know if there is a mechanism by which I can prevent the pdflush daemon from swapping the pages that my filesystem refers to?? What limited knowledge I have on pdflush daemon, I guess that the daemon searches for the dirty inodes in the list maintained per super block. So if I remove all the inodes from this list (or not add any of my inodes I create in the list), will I be able to prevent the daemon from flushing data from my file system?

3 3

Regarding the basic use of perf top command
by ankur dwivedi July 14, 2011

July 14, 2011

Hi, My question is regarding the basic use of perf top command in linux. The following is the snap shot of running a "*perf top -e cache-misses*" command from console: ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- PerfTop: 76* irqs/sec* kernel:14.5% exact: 0.0% [1000Hz cache-misses], (all, 12 CPUs) ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- *samples* *pcnt* function DSO _______ _____ _______________________ ____________________________________ 48.00 36.9% get_page_from_freelist [kernel.kallsyms] 12.00 9.2% JS_CallTracer /usr/lib/xulrunner-1.9.2/libmozjs.so 12.00 9.2% JS_TraceChildren /usr/lib/xulrunner-1.9.2/libmozjs.so 8.00 6.2% dso__find_symbol /usr/bin/perf 7.00 5.4% native_flush_tlb_single [kernel.kallsyms] 6.00 4.6% _int_malloc /lib/libc-2.12.90.so What i want to know is the significance of different parameters here which are the following: 1. irqs/sec 2. samples 3. pcnt It also shows a value of 1000Hz. what is the meaning of that. I searched regarding these in the internet, but was unable to find a good explanation Thanks Ankur

2 2