September 2015 - kernelnewbies@kernelnewbies.org

How to handle Hotplug with UIO userspace driver
by Mandeep Sandhu Sept. 28, 2017

Sept. 28, 2017

Hi All, We're working on a custom PCIe based hardware, which can be hotplugged into the system. We're using the UIO approach of having a minimal kernel driver which basically handles the interrupt and provides a mmap of the device memory to userspace and userspace driver which mmap's the provided memory and manages the device. We're using systemd and udev rules to spawn off a user process whenever the hardware is plugged into the system. Now my question is, how can we gracefully inform the userspace process if someone pulls out the hardware (sitting on a PCI slot). Even with full hotplug support, there can always be a case of "surprise removal", where someone yanks the board without going through the whole "attention button" routine. While the kernel driver has no problem with hot-removal, I see that if I "unregister" our UIO driver in response to the removal, then the mapping for the hardware's physical memory would become "invalid" (done by UIO on unregsiter). This could possibly cause the user-space process to crash due to illegal memory access. Is there a "standard" way to handle this scenario, i.e for hotpluggable hardware using UIO? Basically, we'd want to "unregister" the UIO device only after the userspace process is informed of the removal so that it could unmap/close the device and exit gracefully. We could try doing it using some systemd trick, where we put a dependency on one of the UIO created files (/dev/uioX), so that when the file is destroyed, systemd could kill/shutdown the process. Although, there's no guarantee that the userspace process won't try to access the mmaped space during that interval, so this isn't a reliable solution either. Any hints, thoughts? Thanks., -mandeep

3 7

Re: [PATCH] 8250_pci: Prevent Exar/RTD Boards from binding.
by Greg KH Nov. 12, 2015

Nov. 12, 2015

On Fri, Sep 25, 2015 at 07:08:32PM +0000, Rob Groner wrote: > > -----Original Message----- > > From: Greg KH [mailto:greg@kroah.com] > > Sent: Friday, September 25, 2015 2:37 PM > > To: Rob Groner <rgroner(a)RTD.com> > > Cc: Valdis.Kletnieks(a)vt.edu; kernelnewbies(a)kernelnewbies.org > > Subject: Re: [PATCH] 8250_pci: Prevent Exar/RTD Boards from binding. > > > > On Fri, Sep 25, 2015 at 05:37:03PM +0000, Rob Groner wrote: > > > > > > > -----Original Message----- > > > > From: Valdis.Kletnieks(a)vt.edu [mailto:Valdis.Kletnieks@vt.edu] > > > > Sent: Friday, September 25, 2015 12:48 PM > > > > To: Rob Groner <rgroner(a)RTD.com> > > > > Cc: kernelnewbies(a)kernelnewbies.org > > > > Subject: Re: [PATCH] 8250_pci: Prevent Exar/RTD Boards from binding. > > > > > > > > On Fri, 25 Sep 2015 11:46:29 -0400, Rob Groner said: > > > > > Serial boards made by RTD using the Exar XR17V358 chip rely on the > > > > > extra capabilities of the Exar-provided driver to allow > > > > > configuration of the board. When support for the Exar chip was > > > > > added to the kernel 8250_pci driver, this then prevented easy use > > > > > of the board by customers for anything other than standard serial usage > > in RS232 mode. > > > > > > > > Was it your intent to also prevent the use of this board in standard > > > > serial usage in RS232 mode (which I'd expect is the most common use > > case)? > > > > > > That is a byproduct of giving the non-average user the ability to > > > reconfigure their board. This will basically move us back to pre-3.8, > > > where the customer would simply have to insmod the provided Exar > > > driver. The small inconvenience to that more common user seems (to us > > > in Tech Support) outweighed by the much greater inconvenience to the > > > user who wants to reconfigure. > > > > Where is the exar driver, in the kernel already? > > > > confused, > > I'm sorry for the confusion. Let me summup: > > We produce a serial port board that uses the Exar XR17V358 chip. The board features a jumperless configuration so that to change the board from RS232 to RS422/RS485, you use the GPIO available on the Exar chip, via the Exar driver. That driver is provided by Exar (from their website, and repackaged on our website and with the board). > > Recently, we began to hear from customers who purchased the board but could not get the driver to find the board (and thus could not reconfigure it, nor use the non-standard high baud rates the chip is capable of). We discovered that in 3.8, support for the Exar chip was added to the 8250_pci driver, thus binding it to the kernel. > > Until (and probably if) Exar decides to submit their driver to the kernel, then it leaves us with a problem that we didn't have prior to 3.8...namely that the board won't do what it is advertised to do unless the customer rebuilds the kernel (that is the only supported workaround from Exar). The only other workaround we know of (unbind) has met with mixed success which I won't go into unless you want me to, and is already resisted by some customers. > > The goal of this patch is to get to a point where a customer can install Linux and have full use of this RTD board (using the driver Exar/RTD provides). No one who has an RTD board is going to feel this is an inconvenience. Can you point me at the driver and I'll be glad to add it to the kernel so that the proper driver will bind to the device and this will not be an issue for users? thanks, greg k-h

4 14

(no subject)
by Shraddha Barke Oct. 28, 2015

Oct. 28, 2015

I updated my local kernel repository with all recent commits using the following commands- git checkout staging-next git pull After that when I try to compile, I'm getting an error scripts/sign-file.c:20:25: fatal error: openssl/bio.h: No such file or directory #include <openssl/bio.h> ^ compilation terminated. scripts/Makefile.host:91: recipe for target 'scripts/sign-file' failed make[1]: *** [scripts/sign-file] Error 1 Makefile:545: recipe for target 'scripts' failed make: *** [scripts] Error 2 What should I do to fix it? Thanks in advance.

3 3

SHA-1 hash calculate in Kernel.
by lx Oct. 22, 2015

Oct. 22, 2015

hi all: I built a module for calculate the SHA-1. The code is: ################## #include <linux/init.h> #include <linux/module.h> #include <linux/kernel.h> #include <linux/crypto.h> #include <linux/err.h> #include <linux/scatterlist.h> MODULE_LICENSE("Dual BSD/GPL"); #define SHA1_LENGTH 20 static int hello_init(void) { /* * http://lxr.oss.org.cn/source/fs/ecryptfs/crypto.c?v=2.6.30 */ struct scatterlist sg; struct hash_desc desc; //way 1 /* char *plaintext = "c"; size_t len = strlen(plaintext); */ //way 2 /* char *plaintext = kmalloc(sizeof(char), GFP_KERNEL); plaintext = "c"; size_t len = 1; */ // way 3. /* char plaintext[1] = {'c'}; size_t len = 1; */ // way 4. /* char *plaintext = (char *)__get_free_page(GFP_KERNEL); memcpy(plaintext, "c", 1); size_t len = 1; */ int rc = 0; int i; char hashtext[SHA1_LENGTH]; memset(hashtext, 0x00, SHA1_LENGTH); printk(KERN_INFO "sha1: %s\n", __FUNCTION__); sg_init_one(&sg, plaintext, len); desc.tfm = crypto_alloc_hash("sha1", 0, CRYPTO_ALG_ASYNC); desc.flags = 0; rc = crypto_hash_init(&desc); if (rc) { printk(KERN_ERR "%s: Error initializing crypto hash; rc = [%d]\n", __func__, rc); goto out; } rc = crypto_hash_update(&desc, &sg, len); if (rc) { printk(KERN_ERR "%s: Error updating crypto hash; rc = [%d]\n", __func__, rc); goto out; } rc = crypto_hash_final(&desc, hashtext); if (rc) { printk(KERN_ERR "%s: Error finalizing crypto hash; rc = [%d]\n", __func__, rc); goto out; } crypto_free_hash(desc.tfm); for (i = 0; i < 20; i++) { printk(KERN_INFO "%02x-%d\n", hashtext[i]&0xff, i); } out: printk(KERN_INFO "end\n"); return rc; } static void hello_exit(void) { printk(KERN_ALERT "Goodbye, cruel world\n"); } module_init(hello_init); module_exit(hello_exit); ################## I can just get the right result by way 3 and 4, In way 1 and 2 , I get the wrong result, I check the result by link: http://www.xorbin.com/tools/sha1-hash-calculator . Please tell me why? Thank you.

4 4

Re: Help
by Mulyadi Santosa Oct. 5, 2015

Oct. 5, 2015

On Fri, Sep 25, 2015 at 10:14 PM, Prem Kumar <prem.it.kumar(a)gmail.com> wrote: > Dear Mulyadi, > > Sorry for sending direct email. I am at a deadlock pulling my hairs and > can't seem to figure out ( > http://lists.kernelnewbies.org/pipermail/kernelnewbies/2015-September/01511…) > how to reclaim the AnonHugePage allocations when only system services are > running on them. Reboot is an option but too many nodes and this happening > frequently i need to know an alternate way out. > > Any help is greatly appreciated. > Regards, > Prem > > > Dear Prem Sorry that I can not promptly answer any question these days. My quick conclusion is that, since you said you alocate huge page above, then I guess huge page has different reclaim policy, that's why it might not be released when your application terminated. maybe you forgot to explicitly release them? All in all, I appreciate you ask my help, but next time please make sure that person is inviting you for direct discussion first, so you won't be considered as impolite. -- regards, Mulyadi Santosa Freelance Linux trainer and consultant blog: the-hydra.blogspot.com training: mulyaditraining.blogspot.com

2 1

Unset LOCKDEP and TRACE_IRQFLAGS_SUPPORT
by vibnwis Oct. 5, 2015

Oct. 5, 2015

Hi all, I need to unset default value of LOCKDEP(=y) and TRACE_IRQFLAGS_SUPPORT(=y) to both "n". Having asked around and had tried for at least 3 days, seriously I still couldn't get both items configured as "n" Perhaps someone can help me on this item first. For instance, Symbol TRACE_IRQFLAGS description below, to set it, it will need the logics below: PROVE_LOCKING [=y] && DEBUG_KERNEL [=y] && TRACE_IRQFLAGS_SUPPORT [=y] && STACKTRACE_SUPPORT [=y] && LOCKDEP_SUPPORT [=y] || IRQSOFF_TRACER [=n] && TRACING_SUPPORT [=y] && FTRACE To unset it, I could either set DEBUG_KERNEL=n or set TRACING_SUPPORT=n (or FTRACE=n). However, having set it, yet I don't see symbol TRACE_IRQFLAGS is unset. Anything wrong of my interpretation of the logics above? Copied from "/" type TRACE_IRQFLAGS in menuconfig │ Symbol: TRACE_IRQFLAGS [=y] │ │ Type : boolean │ │ Defined at lib/Kconfig.debug:1047 │ │ Selected by: PROVE_LOCKING [=y] && DEBUG_KERNEL [=y] && TRACE_IRQFLAGS_SUPPORT [=y] && STACKTRACE_SUPPORT [=y] && LOCKDEP_SUPPORT [=y] || IRQSOFF_TRACER [=n] && TRACING_SUPPORT [=y] && FTRACE [=n │ │ │ │ │ │ Symbol: TRACE_IRQFLAGS_SUPPORT [=y] │ │ Type : boolean │ │ Defined at arch/arm/Kconfig:174 Appreciate any comments. Thanks in advance. Lim

4 24

Can I send a patch about 00-INDEX?
by Navy Cheng Oct. 2, 2015

Oct. 2, 2015

Hi, I read some document about schedule in Documentation/schedule/ but I find that there are 9 files in 00-INDEX (include itself), however there are 10 files in the Documentation/schedule/* . *completoin.txt* is added in 2015 and 00-INDEX is not include it. Should I send a patch like this? Thank you?

2 1

Accessing pointers inside struct passed as argument to ioctl calls
by Daniel. Sept. 29, 2015

Sept. 29, 2015

Hi all, I have a doubt about using pointers inside structs that are passed (as pointers) to ioctl argument. Since pointers passed from userspace can't be trusted, I need to copy they to kernel before accessing they. In this case I have a pointer inside a struct that is passed to the ioctl call also as pointer. So I need to copy the whole struct to kernel space and only then dereference the pointer. If this is true, two copy_from_user is needed right? Suppose I have a struct like this struct myioctl_arg { char *tx_buf; int tx_siz; } and this ioctl definition #define MYIOCTL_TX _IOWR(MY_MAGIC, MY_BASE, struct myioctl_arg *); At userspace program I do something like this: struct myioctl_arg arg = { .tx_buf = somebuf, .tx_siz = 32 } ioctl(fd, MYIOCTL_TX, &arg); And in my ioclt implementation static ioctl(struct file *fp, unsigned int cmd, unsigned long arg) { struct myioctl_arg karg; char kbuf[100]; ... case MYIOCTL_TX: copy_from_user(&karg, arg, sizeof(karg)); copy_from_user(kbuf, karg.tx_buf, karg.tx_siz); // now kbuf has the same contents of somebuf in userspace ... } I'm in doubt if this is the right way to access the userspace tx_buf. Since the .tx_buf from arg is a userspace pointer, accessible from userspace pointer arg, I can't type arg->tx_buf directly, doing this is dereferencing a userspace pointer inside kernel. So I need to copy the whole struct and dereference from the kernel space copy of it. So this two calls of copy_from_user() is the right way to do it? They are needed at all or is there better way of doing it? Cheers, -- "Do or do not. There is no try" Yoda Master

4 6

Unable to get stmmac driver of Altera Cyclone V to support MTU9000
by Teoh Choon Zone Sept. 28, 2015

Sept. 28, 2015

According to the technical reference manual, Cyclone V does support jumbo frame up to 9.6KB, which I read as it will support MTU up to 9000. But I couldn't get it to work properly with MTU9000. I define working as in I can ping with a packet size of 9000 bytes, e.g "ping <host> -s 9000", the ping process just won't start when I set the MTU anything larger then 8144. FYI, tx is running in store-and-forward mode and rx is running in threshold mode. Anyone know how I can tweak the stmmac driver to support MTU9000?

3 6

Re: Unset LOCKDEP and TRACE_IRQFLAGS_SUPPORT
by Valdis.Kletnieks＠vt.edu Sept. 28, 2015

Sept. 28, 2015

On Mon, 28 Sep 2015 11:42:34 +1300, vibnwis said: > > See the difference? > The kernel I work is for Pandaboard. Hence NVidia will not in the picture. The point is that one explanation gives a lot more confidence and understanding than the other. I'll let you figure out for yourself which the suggestions to turn off TRACE_IRQFLAGS sounds more like....

1 0