Message ID | 20211117154101.38659-1-tz.stoyanov@gmail.com (mailing list archive) |
---|---|
Headers | show |
Series | Introduce configurable ring buffer page size | expand |
On Wed, 17 Nov 2021 17:40:58 +0200 "Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com> wrote: > From: "Tzvetomir Stoyanov (VMware)" <tstoyanov@vmware.com> > > Currently the size of one buffer page is global for all buffers and it > is hard coded to one system page. The patch set introduces configurable > ring buffer page size, per ring buffer. A new user space interface is > introduced, which allows to change the page size of the ftrace buffer, per > ftrace instance. A few of nits. 1. Even the cover letter should have a category ("tracing:"). 2. The category is "tracing:" not "trace:" 3. These are kernel changes and should be sent to lkml and not linux-trace-devel, as that mailing list is for tooling. -- Steve > > Tzvetomir Stoyanov (VMware) (3): > [RFC] trace: Page size per ring buffer > [RFC] trace: Add interface for configuring trace ring buffer size > [WiP] trace: Set new size of the ring buffer page > > include/linux/ring_buffer.h | 5 +- > kernel/trace/ring_buffer.c | 251 +++++++++++++++++++++++++++--------- > kernel/trace/trace.c | 47 +++++++ > kernel/trace/trace_events.c | 71 ++++++++-- > 4 files changed, 304 insertions(+), 70 deletions(-) >
On Wed, Nov 17, 2021 at 7:10 PM Steven Rostedt <rostedt@goodmis.org> wrote: > > On Wed, 17 Nov 2021 17:40:58 +0200 > "Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com> wrote: > > > From: "Tzvetomir Stoyanov (VMware)" <tstoyanov@vmware.com> > > > > Currently the size of one buffer page is global for all buffers and it > > is hard coded to one system page. The patch set introduces configurable > > ring buffer page size, per ring buffer. A new user space interface is > > introduced, which allows to change the page size of the ftrace buffer, per > > ftrace instance. > > A few of nits. > > 1. Even the cover letter should have a category ("tracing:"). > 2. The category is "tracing:" not "trace:" > 3. These are kernel changes and should be sent to lkml and not > linux-trace-devel, as that mailing list is for tooling. > Thanks Steven, I'll send the patch set to lkml when it is ready, it is still a work in progress. I just noticed that I forgot to remove some leftovers from my previous implementation in the third patch. It is still not stable, as the kernel freezes in some cases. > -- Steve > > > > > Tzvetomir Stoyanov (VMware) (3): > > [RFC] trace: Page size per ring buffer > > [RFC] trace: Add interface for configuring trace ring buffer size > > [WiP] trace: Set new size of the ring buffer page > > > > include/linux/ring_buffer.h | 5 +- > > kernel/trace/ring_buffer.c | 251 +++++++++++++++++++++++++++--------- > > kernel/trace/trace.c | 47 +++++++ > > kernel/trace/trace_events.c | 71 ++++++++-- > > 4 files changed, 304 insertions(+), 70 deletions(-) > > >
From: "Tzvetomir Stoyanov (VMware)" <tstoyanov@vmware.com> Currently the size of one buffer page is global for all buffers and it is hard coded to one system page. The patch set introduces configurable ring buffer page size, per ring buffer. A new user space interface is introduced, which allows to change the page size of the ftrace buffer, per ftrace instance. Tzvetomir Stoyanov (VMware) (3): [RFC] trace: Page size per ring buffer [RFC] trace: Add interface for configuring trace ring buffer size [WiP] trace: Set new size of the ring buffer page include/linux/ring_buffer.h | 5 +- kernel/trace/ring_buffer.c | 251 +++++++++++++++++++++++++++--------- kernel/trace/trace.c | 47 +++++++ kernel/trace/trace_events.c | 71 ++++++++-- 4 files changed, 304 insertions(+), 70 deletions(-)