Linux DTrace development list
 help / color / mirror / Atom feed
From: Kris Van Hees <kris.van.hees@oracle.com>
To: Kris Van Hees <kris.van.hees@oracle.com>, dtrace@lists.linux.dev
Cc: "eugene.loh@oracle.com" <eugene.loh@oracle.com>,
	dtrace-devel@oss.oracle.com
Subject: Re: [DTrace-devel] [PATCH 1/2] Clarify how the usdt_prids key is stored on the BPF stack
Date: Wed, 19 Mar 2025 11:18:03 -0400	[thread overview]
Message-ID: <Z9rgKzWMHSqLVQdK@oracle.com> (raw)
In-Reply-To: <Z9rXYvNQxZqODC3o@kvh-deb-bpf.us.oracle.com>

On Wed, Mar 19, 2025 at 10:40:34AM -0400, Kris Van Hees via DTrace-devel wrote:
> On Wed, Feb 19, 2025 at 11:43:49PM -0500, eugene.loh@oracle.com wrote:
> > 
> > While one can access the BPF stack relative to %r9, the whole
> > point of DT_TRAMP_SP_SLOT(0) is to make trampoline code more
> > readable.  So use it.
> > 
> > Signed-off-by: Eugene Loh <eugene.loh@oracle.com>
> 
> Reviewed-by: Kris Van Hees <kris.van.hees@oracle.com>

Still applies but see below...

> > ---
> >  libdtrace/dt_prov_uprobe.c | 21 +++++++--------------
> >  1 file changed, 7 insertions(+), 14 deletions(-)
> > 
> > diff --git a/libdtrace/dt_prov_uprobe.c b/libdtrace/dt_prov_uprobe.c
> > index 5d9f74244..f1323cc31 100644
> > --- a/libdtrace/dt_prov_uprobe.c
> > +++ b/libdtrace/dt_prov_uprobe.c
> > @@ -1015,22 +1015,15 @@ static int trampoline(dt_pcb_t *pcb, uint_t exitlbl)
> >  	emit(dlp,  BPF_ALU64_IMM(BPF_RSH, BPF_REG_0, 32));
> >  
> >  	/*
> > -	 * Look up in the BPF 'usdt_prids' map.  Space for the look-up key
> > -	 * will be used on the BPF stack:
> > -	 *
> > -	 *     offset                                       value
> > -	 *
> > -	 *     -sizeof(usdt_prids_map_key_t)                pid (in %r0)
> > -	 *
> > -	 *     -sizeof(usdt_prids_map_key_t) + sizeof(pid_t)
> > -	 *     ==
> > -	 *     -sizeof(dtrace_id_t)                         underlying-probe prid
> > +	 * Look up in the BPF 'usdt_prids' map.  The key should fit into
> > +	 * trampoline stack slot 0.
> >  	 */
> > -	emit(dlp,  BPF_STORE(BPF_W, BPF_REG_9, (int)(-sizeof(usdt_prids_map_key_t)), BPF_REG_0));
> > -	emit(dlp,  BPF_STORE_IMM(BPF_W, BPF_REG_9, (int)(-sizeof(dtrace_id_t)), uprp->desc->id));
> > +	assert(sizeof(usdt_prids_map_key_t) <= DT_STK_SLOT_SZ);
> > +	emit(dlp,  BPF_STORE(BPF_W, BPF_REG_FP, DT_TRAMP_SP_SLOT(0), BPF_REG_0));
> > +	emit(dlp,  BPF_STORE_IMM(BPF_W, BPF_REG_FP, DT_TRAMP_SP_SLOT(0) + sizeof(pid_t), uprp->desc->id));

I get a compiler warning here:

libdtrace/dt_prov_uprobe.c: In function â?~trampolineâ?T:
include/bpf_asm.h:119:24: warning: overflow in conversion from â?~long unsigned intâ?T to â?~short intâ?T changes value from â?~18446744073709551524â?T to â?~-92â?T [-Woverflo]
  119 |                 .off = (ofs),                                           \
      |                        ^
libdtrace/dt_as.h:42:69: note: in definition of macro â?~emitleâ?T
   42 |                  dt_irnode_t *dip = dt_cg_node_alloc((lbl), (instr)); \
      |                                                              ^~~~~

libdtrace/dt_prov_uprobe.c:1013:9: note: in expansion of macro â?~emitâ?T
 1013 |         emit(dlp,  BPF_STORE_IMM(BPF_W, BPF_REG_FP, DT_TRAMP_SP_SLOT(0) + sizeof(pid_t), uprp->desc->id));
      |         ^~~~
libdtrace/dt_prov_uprobe.c:1013:20: note: in expansion of macro â?~BPF_STORE_IMMâ?T
 1013 |         emit(dlp,  BPF_STORE_IMM(BPF_W, BPF_REG_FP, DT_TRAMP_SP_SLOT(0) + sizeof(pid_t), uprp->desc->id));

You need a (int) cast for sizeof(pid_t) similar to the casts that were in
the code before.  I'll add it in as I merge.

> >  	dt_cg_xsetx(dlp, usdt_prids, DT_LBL_NONE, BPF_REG_1, usdt_prids->di_id);
> > -	emit(dlp,  BPF_MOV_REG(BPF_REG_2, BPF_REG_9));
> > -	emit(dlp,  BPF_ALU64_IMM(BPF_ADD, BPF_REG_2, (int)(-sizeof(usdt_prids_map_key_t))));
> > +	emit(dlp,  BPF_MOV_REG(BPF_REG_2, BPF_REG_FP));
> > +	emit(dlp,  BPF_ALU64_IMM(BPF_ADD, BPF_REG_2, DT_TRAMP_SP_SLOT(0)));
> >  	emit(dlp,  BPF_CALL_HELPER(BPF_FUNC_map_lookup_elem));
> >  	emit(dlp,  BPF_BRANCH_IMM(BPF_JEQ, BPF_REG_0, 0, lbl_exit));
> >  
> > -- 
> > 2.43.5
> > 
> > 
> 
> _______________________________________________
> DTrace-devel mailing list
> DTrace-devel@oss.oracle.com
> https://oss.oracle.com/mailman/listinfo/dtrace-devel

  parent reply	other threads:[~2025-03-19 15:18 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-02-20  4:43 [PATCH 1/2] Clarify how the usdt_prids key is stored on the BPF stack eugene.loh
2025-02-20  4:43 ` [PATCH 2/2] Extend the USDT bit mask to multiple words eugene.loh
     [not found] ` <Z9rXYvNQxZqODC3o@kvh-deb-bpf.us.oracle.com>
2025-03-19 15:18   ` Kris Van Hees [this message]
2025-03-19 16:30     ` [DTrace-devel] [PATCH 1/2] Clarify how the usdt_prids key is stored on the BPF stack Eugene Loh
  -- strict thread matches above, loose matches on Subject: below --
2025-03-19 14:47 Kris Van Hees

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=Z9rgKzWMHSqLVQdK@oracle.com \
    --to=kris.van.hees@oracle.com \
    --cc=dtrace-devel@oss.oracle.com \
    --cc=dtrace@lists.linux.dev \
    --cc=eugene.loh@oracle.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox