Re: [PATCH] acpi/nfit: Fix bus command validation

2019-02-21 Thread Johannes Thumshirn
[+CC dvyukov ] On 20/02/2019 18:21, Johannes Thumshirn wrote: > On 20/02/2019 17:15, Dan Williams wrote:> I wouldn't be opposed to > syzkaller fuzzing the nvdimm-ioctl path. > As a heads up, I've started adding the ioctl() definitions to syzcaller. > Just so we don't duplicate any efforts. So

Re: [PATCH] acpi/nfit: Fix bus command validation

2019-02-20 Thread Johannes Thumshirn
On 20/02/2019 17:15, Dan Williams wrote:> I wouldn't be opposed to syzkaller fuzzing the nvdimm-ioctl path. As a heads up, I've started adding the ioctl() definitions to syzcaller. Just so we don't duplicate any efforts. Byte, Johannes -- Johannes Thumshirn

Re: [PATCH] acpi/nfit: Fix bus command validation

2019-02-20 Thread Dan Williams
On Wed, Feb 20, 2019 at 12:30 AM Johannes Thumshirn wrote: > > On 20/02/2019 03:58, Dan Williams wrote: > [...] > > >> > >> Tricky code path, eh? > > > > ioctl path, number one source of bugs / thrash in this subsystem. 2nd > > place, ARS. > > Possibly unpopular idea, but should we maybe teach

Re: [PATCH] acpi/nfit: Fix bus command validation

2019-02-20 Thread Johannes Thumshirn
On 20/02/2019 03:58, Dan Williams wrote: [...] >> >> Tricky code path, eh? > > ioctl path, number one source of bugs / thrash in this subsystem. 2nd > place, ARS. Possibly unpopular idea, but should we maybe teach trinity/syzcaller about these ioctl()s? Better we find the bugs in a QA like

Re: [PATCH] acpi/nfit: Fix bus command validation

2019-02-19 Thread Dan Williams
On Tue, Feb 19, 2019 at 5:57 PM Jeff Moyer wrote: > > Dan Williams writes: > > > Commit 11189c1089da "acpi/nfit: Fix command-supported detection" broke > > ND_CMD_CALL for bus-level commands. The "func = cmd" assumption is only > > valid for: > > > > ND_CMD_ARS_CAP > > ND_CMD_ARS_START >

Re: [PATCH] acpi/nfit: Fix bus command validation

2019-02-19 Thread Jeff Moyer
Dan Williams writes: > Commit 11189c1089da "acpi/nfit: Fix command-supported detection" broke > ND_CMD_CALL for bus-level commands. The "func = cmd" assumption is only > valid for: > > ND_CMD_ARS_CAP > ND_CMD_ARS_START > ND_CMD_ARS_STATUS > ND_CMD_CLEAR_ERROR > > The function

Re: [PATCH] acpi/nfit: Fix bus command validation

2019-02-07 Thread Verma, Vishal L
On Thu, 2019-02-07 at 15:57 -0800, Dan Williams wrote: > Commit 11189c1089da "acpi/nfit: Fix command-supported detection" broke > ND_CMD_CALL for bus-level commands. The "func = cmd" assumption is only > valid for: > > ND_CMD_ARS_CAP > ND_CMD_ARS_START > ND_CMD_ARS_STATUS >

[PATCH] acpi/nfit: Fix bus command validation

2019-02-07 Thread Dan Williams
Commit 11189c1089da "acpi/nfit: Fix command-supported detection" broke ND_CMD_CALL for bus-level commands. The "func = cmd" assumption is only valid for: ND_CMD_ARS_CAP ND_CMD_ARS_START ND_CMD_ARS_STATUS ND_CMD_CLEAR_ERROR The function number otherwise needs to be pulled from the